▶ 0:14:28Good morning everybody. The committee will come to order. Welcome everybody of the committee back. We appreciate everybody being back this morning for I think it's going to be an absolutely exciting hearing. And I will recognize myself for five minutes for an opening statement. And I want to thank our witnesses for being here and traveling long distance to be here. Um, this kind of came from a library of congress this hearing from a presentation that Dr.
▶ 0:14:54Smith put on his book Genesis that he wrote with Henry Kissinger, Henry Kissinger's last book. And I walked away uh thinking we needed to have the entire Congress hear your presentation. And so we're doing it with the entire energy and commerce committee because I think an author always wants to know when they write a book, will somebody read my book? And then if they read the book, then will it have an impact?
▶ 0:15:14Well, today you and all the witnesses are before the full energy commerce committee and our our dear colleague and and the dear husband of of our colleague Debbie Dingle used to say that if it's moving it's energy if it stops his commerce something to that effect. So we have a lot of jurisdiction. I say it takes energy to move commerce. I can't improve on Chairman Dingle but that's my version of it. And so we're having a full hearing.
▶ 0:15:38We typically do this in subcommittee but but this touches all jurisdictions and I think everybody needs to hear it. If you think about uh it's going to take enormous energy to beat China to AI we in doing that we have to protect the environment our telecom and privacy through our commerce and telecom committees will be dealing with this and AI has particularly healthc care applications so it touches all of our jurisdiction and and Dr. Mr.
▶ 0:16:05Smith, when I walked away from the Library of Congress and I read your book, uh, it gave me a sense of mission in the mission that a direction I want to take this committee in the time that I'm chairman. And to sum up what you said, it's the US versus China and who will win the war for AI. And it it essentially I walked away this is as important as the dollar being the reserve currency in the world. It's that important. That's what's before us.
▶ 0:16:32and and we what I what I hear from people in this space is that we have the brain power and we have the capital. What we need is the energy and the correct regulatory framework. Um and we have an example of what not to do and I believe you said ch Europe in your presentation Europe has chosen not to grow. So we can't look there as an example.
▶ 0:16:53We have to work through it our and and in Europe is the regulatory framework their energy framework and the regulatory framework some of their regulatory framework written specifically just to disadvantage American companies has made them non-competitive and the Europe and the US had a similar size economy in 2008 and I've read that our economy is up about 80% larger so what do we need to do and the reason we want to do a full committee is that we have to have broad consensus on how we work together.
▶ 0:17:23It has to be Democrat and Republican. People who um tell me they invest, it's tough to invest based on congressional cycles or presidential cycles that the rules are going to change every two to four years. And so what I would like to hopefully we could do in this committee is come up with a a regulatory framework and a energy policy that we can all or most of us can agree on at least build a broad consensus on how we develop massive amounts of energy while protecting our environment. And uh Dr.
▶ 0:17:53Schmidt, you said all energy resources are needed and then AI will develop solutions to deal with climate change. And so Microsoft, to put this in perspective, Microsoft data center can use as much power as a city of Seattle is what I've been told. And so in the regulation side of it, we have to protect our privacy. Yesterday we had a hearing or bills yesterday on on child children's privacy and children's safety. And we have to protect our privacy. I think all of us want our privacy protected.
▶ 0:18:24We can't do it in a heavy-handed way that stifles innovation. And as I said, we have to look at our friends across the Atlantic. But I think we need to more intently look across the Pacific to a nation determined to win. China has specifically said they are going to win the war on AI. And we are taking up the challenge to prove to them that the American entrepreneur and the American intellect will win the war on AI.
▶ 0:18:54But they have to have the energy and the regulatory environment to do so. So if this committee gets it right, this committee gets it right, America will win. It may win if otherwise, but we need to be there to to to make that happen. And if you look at what if China wins, we just had a hearing an oversight that a medical device from China had an embedded URL to the University of Beijing. So why did that mean a medical device?
▶ 0:19:24Because we know they're using everything they can, everything they can to get information they need on us. So we must win, we will win, and for the sake of the world, we have to win. And I determined through this hearing to the beginning that all of us will work together because all of us are dedicated to winning. And I will yield back and recognize the ranking member for five minutes for an opening statement. Thank you, Mr. Chairman.
▶ 0:19:51Under normal circumstances, today's hearing would be a bipartisan conversation on ensuring America continues to lead the race on artificial intelligence or AI. However, these are not normal times. President Trump is single-handedly destroying our economy since he unnecessarily instigated a global trade war. Our markets are in turmoil. Americans retirement savings is in freef fall and prices for everyday everyday goods are spiking.
▶ 0:20:18In fact, Trump's tariffs are the largest middle class tax increase in at least 50 years on hardworking American families. And our efforts to continue to lead the global race on AI innovation are seriously threatened when Trump has just spiked the price on materials. we need to compete such as steel, aluminum, and chips. Instead of winning the future, Trump's economic turmoil could send America's tech leadership into a tail spin.
▶ 0:20:43There's no doubt that the daily chaos and uncertainty that Trump is creating is not good for American business or for the American people. Despite the unwillingness of the president and Republicans to acknowledge any of the harm their actions are having on American families, I want to address the topic of today's hearing because it is so important. As we have heard in every energy subcommittee hearing this year, increased energy demand is coming, largely powered by data centers fueling artificial intelligence tools.
▶ 0:21:11And I firmly believe that this increased demand can be a good thing, but it must be managed responsibly. We must make sure that AIdriven energy demand increases don't make electricity unaffordable or unreliable for American families. We must also make sure that consumers aren't stuck varying the costs for infrastructure investments made necessary by private companies. And we must get a better understanding of just how much energy demand will increase in the coming years. The committee needs to be talking about all these things.
▶ 0:21:41But instead, this week, House Republicans are poised to vote on a budget resolution that would set the stage to repeal the energy tax credits incentivizing well over 90% of the electricity generation poised to come onto the grid. The Trump administration and Elon Musk Doge minions are also putting together a secret list of grants and loans that they want to cancel that would modernize our electric grid and build new energy generation.
▶ 0:22:06Meanwhile, yesterday afternoon, Trump signed a several executive orders to allow polluting coal plants to set for retirement to continue to operate, increasing prices and health risks for American families. And just last month, during his speech to the joint session of Congress, Trump threatened to repeal the Chips and Science Act, which invested $52 billion to ensure more semiconductors are produced right here in the US. Semiconductors are critical to the advancement of AI.
▶ 0:22:33But right now, the overwhelming majority are produced outside the United States, and the Chips and Science Act is boosting production of chips here. And now, Trump wants to repeal the law. So, Republicans constantly talk about winning the AI race, but the actions they're taking make it appear as if they're purposely trying to lose that race to China. And we should also discuss the tremendous effects AI will have on our everyday lives.
▶ 0:22:56We've seen an explosion of AI systems and tools that have been trained on massive amounts of Americans personal information without our knowledge and consent. Right now, sufficient guardrails do not exist to protect Americans and our data from harmful AI systems that violate our privacy, provide false information, or make unjustifiable discriminatory decisions.
▶ 0:23:17Because many of these systems are trained on massive amounts of data that big tech has collected on all of us, the lack of nationwide protections around what data companies can collect, use, and sell to train these AI systems should concern every American. Clearly defined privacy and data security rules are critical to protect consumers from existing harmful data collection practices and to safeguard them from the growing privacy threat that AI models pose.
▶ 0:23:44So I strongly believe that the bedrock of any AI regulation must be privacy legislation built on the principle of limiting the amount of consumer data collected, used, and shared. It is the best way to address the aggressive and abusive data collection practices of big tech and data brokers. ensure our children's sensitive information information is protected online and put consumers back in control of their data.
▶ 0:24:07So, I look forward to hearing from today's witnesses and intend to continue to focus on developing policies that harness the transformation power of AI while safeguarding the rights and well-being of all Americans. And with that, I yield back. Thank you, Mr. Chairman. Thank you. The gentleman yields back and we now conclude with member opening statements. The chair would like to remind members that pursuant to the committee rules, all members opening statements will be made part of the record.
▶ 0:24:32I'd also remind members that once we get to the five-minute questioning, we will have to strictly enforce that. Uh we have time constraints with some of our witnesses and I want everybody to have the chance to ask their questions. I would like to thank the the witnesses for being here today. It it is and you're taking time to testify before this committee is greatly appreciated. you will have the opportunity to give an opening statement followed by a round of questions for members and our witness.
▶ 0:24:59I'll read the witnesses and I'll call on you individually to to read for your opening statement. So, first we have Dr. Eric Schmidt, chair of the special competitive studies project. Dr. Schmidt previously served as a chief executive officer and chairman of Google. In addition to serving executive chairman and technical adviser, his time at Google would turn the company into the global tech giant we know it today.
▶ 0:25:20In 2021, he founded the nine part nonpartisan special competitive studies project to strengthen America's long-term competitiveness regarding AI and America's future and also the author as we said in a library of congress spokesman of the book Genesis he wrote with Dr. Kissinger. So thank you for being here. Dr. Manish Batia, Mr. Manish Batia, executive vice president of global operations with Micron Technology. Mr.
▶ 0:25:45Batia has been with Micron since 2017 and has 25 years of engineering and operations experience. He has previously held positions at West Western Digital Corporation, Sandis Corporation, and Matrix Semiconductor to name just a few. The Honorable David Turk, a visiting fellow with the Center on Global Energy Policy at Columbia University School of International Public Affairs. Mr. Mr. Turk served as the deputy secretary of the US Department of Energy during the Biden administration.
▶ 0:26:14Before his time as deputy secretary, Mr. Tur Turk spent several years at the International Energy Agency. Thank you for being here as well. And Mr. Alexander Wang, the founder and chief executive officer of Scale AI. Mr. Wang founded Scale AI as a 19-year-old student in MIT focusing on the concept of humanity first artificial intelligence. Currently, Scale AI has a team of over 900 and is valued at nearly 14 billion dollars.
▶ 0:26:41At 24, he is the youngest self-made billionaire in the world. So, I thank you all for being here today and I will call on each of you and I'll begin with Dr. Schmidt. You have five minutes for your opening statement. Thank you. And you'll see I'm before you get started there. You'll have a green light and when it gets to four minutes, I think a light turns yellow. So, it'll kind of give you a warning in front of you. you have a minute and when it turns red it'll be wrap it up so we can make sure we get all our questions in. So Dr. Smith your five minutes you're recognized. Um thank you Mr.
▶ 0:27:11Chairman and thank you ranking member. Thank you all for being here. This is incredibly important. I'm here to tell you that I honestly believe that the AI revolution is underhyped. And here's why. The arrival of this new intelligence will profoundly change our country and the world in ways we cannot fully understand. And none of us including myself and frankly anyone in this room is prepared for the implications of this.
▶ 0:27:38What's happening at the moment in our industry is that we're very very quickly for example developing developing AI programmers and these AI programmers will replace uh traditional software programmers. We're building in the next year AI mathematicians that are as good as the top level graduate students in math. This is happening very quickly. You can look at this in a number of the uh products today. Hey, you think of AI as chat GBT, but what it really is is a reasoning and planning system that we've never seen before.
▶ 0:28:09The implication of this is profound. Um, in terms of the way the algorithms work, they're going to need a lot more computation than we've ever had. They're going to need a lot more energy, and I'll talk about that. What does the industry need? We need high skills immigration. We talk to you about this every day. uh light touch regulation around cyber and bio threats. We can talk about that and most importantly we need the energy and the numbers are profound.
▶ 0:28:36Um what we need from you if if I may say that directly is need energy in all forms um renewable non-renewable whatever it needs to be there and it needs to be quickly. I and others are investing in things like fusion which are incredible but they're not going to arrive soon enough for the need. And I'll frame this at the end by my comments about China. So people are planning 10 gigawatt data centers.
▶ 0:29:02Now just to do the translation, an average nuclear power plant in the United States is one gigawatt. How many gig how many nuclear power plants can we make in one year where we're planning this 10 gawatt data center? Gives you a sense of how big this crisis is. Um many people think that the demand in of energy part that our industry takes will go from 3% to 99% of total generation.
▶ 0:29:30One of the estimates that I think is most likely is that data centers will require an additional 29 gawatts of power by 2027 and 67 more gigawatts by 2030. Gives you a sense of the scale that we're talking. These things are industrial at a scale I have never seen in my life.
▶ 0:29:48the in the terms of energy planning the current model is mostly natural gas peaker plants plus renewables and that's probably going to be the path we're going to have to follow right to get there and u for all the reasons that you can imagine we have a bunch of regulatory issues around fixing the the energy grid. It takes on average 18 years to get the power transmissions and so forth to put these things in place. We need to find federal ways to preempt that and make it happen faster in order to deal with the needs.
▶ 0:30:18Many of these data centers by the way are in the heartland. Um they have a huge economic impact positively on areas that typically do not have the kind of growth that they would like. Now why is this all important? When you build these systems, you have intelligence in the computer and then eventually human level intelligence. Some people think it's within three to four years. Then after that you have something called super intelligence.
▶ 0:30:46And super intelligence is the intelligence that's higher than of humans. We believe as an industry that this could occur within a decade. It is crucial that America get there first. What is China doing? They're leading in something open source. They are very close behind us. You all have done a great job in doing chip restrictions and things like that to try to slow them down. They're clever and they're smart.
▶ 0:31:11They have industrial programs, huge grants going into these companies and they're weaponizing up in the sense of competition. If you look at Deep Seek, Deepseek showed up, right? Nobody expected this. It turns out it's on par with some of the top models. Welcome. China has arrived in the competition. What would happen if China beat us? Let's think about it. The path to intelligence, that's superhuman intelligence. Think of the national security implications of that competition.
▶ 0:31:41This is why I believe and I'll say directly to you that although everyone is concerned about Taiwan, I'm much more concerned about this because if they come to super intelligence, this strong form of intelligence first, it changes the balance of power globally in ways that we have no way of understanding, predicting, or dealing with. Thank you, Mr. Chairman. Thank you. Thank you for that sober assessment why we wanted to have this hearing. We appreciate that very much. Uh now we'll recognize Mr. Batia, you have your five minutes for for your opening statement.
▶ 0:32:13Thank you, Mr. Chairman. Ranking member Palone and members of the committee. My name is Manish Batya and I serve as executive vice president of global operations at Micron. Micron was founded in 1978 in Boise, Idaho, and over the last several decades has become one of the world's most innovative companies with more than 58,000 US patents granted.
▶ 0:32:37And Micron is America's only manufacturer of memory chips and the only US semiconductor company with worldwide technology leadership today. Micron is truly a national treasure. Micron has become fundamental to America's economic competitiveness because our fabs manufacture the world's most advanced memory chips and are at the heart of the AI revolution. For each AI chip that Nvidia sells, there are 96 high bandwidth memory chips integrated with it.
▶ 0:33:06Without our chips, there simply is no AI. Micron is the only company planning to invest more than a hundred billion dollars over the next 20 years to build leading edge memory fabs here in the United States. These investments will power America's AI leadership. They'll serve domestic demand for other industries and drive US semiconductor exports.
▶ 0:33:28Our investments are projected to create 11,000 high-paying direct micron jobs, 9,000 construction jobs, and ultimately between direct and indirect 80,000 new jobs created across our expansions planned in Idaho, New York, and Virginia. The president and Congress have made clear that the United States needs to continue to lead on AI and increase domestic manufacturing. The success of our investments will keep the US at the forefront of the AI revolution.
▶ 0:33:58strengthen the economy and make America secure. To make our historic US investments, we need reliable and affordable energy. One of the most important factors that made upstate New York and Boise, Idaho attractive for our planned investments is reliable, lowcost power.
▶ 0:34:16And in Virginia, where we've been operating for two decades, grid reliability has been critical to our Each of these full-scale fabs built here will run 24 by7 by 365 days a year and consume at full buildout about 400 megawatts of power. By 2040, we expect our US energy demands to reach 2 gawatt.
▶ 0:34:40This demand comes from a variety of highly complex manufacturing process steps, including using extreme ultraviolet lasers to create advanced nanocale features on our chips. Beyond scale, we also need power to be reliable.
▶ 0:34:55Even fractions of a seconds of power loss or even just power sag or droop forces us to reset equipment, check for inconsistencies and deviations in the material and um ultimately can cost tens or even hundreds of millions of dollars.
▶ 0:35:12Reliable power is critical to our US Historically, the United States has maintained low electricity prices due to the abundance of energy resources and it's all of the above approach from oil and natural gas to solar and nuclear. This was a bright spot for Micron as we built here at home and is one of America's key competitive advantages in manufacturing.
▶ 0:35:36However, after years of matched supply and demand, we're now seeing significant electricity demand growth and supply may struggle to keep pace. By one estimate, US electricity demand could rise by 128 gigab gigawatts, more than 15% over the next 5 years alone. This risks the United States losing leadership in AI and in the technologies that enable it.
▶ 0:36:00Meeting this energy demand means the federal government needs to take an all of the- above approach and cut through red tape to bring generating projects to life. We also need to invest in energy equipment and supply chains. When I visited the Idaho National Lab last month to discuss their cutting edge work on advanced nuclear technologies, it became clear how much investment is needed in uranium fuel supply chains and other new technologies.
▶ 0:36:24Beyond generating capacity and energy supply chains, we need to ensure the US transmission infrastructure is fit for the 21st century. Without new and updated transmission infrastructure, new generation won't deliver won't be able to be delivered to customers like us. This is why permitting reform to accelerate transmission infrastructure is so important. Taking a step back and looking at manufacturing and AI more broadly, this also means continued investment in manufacturers that enable the AI revolution.
▶ 0:36:55Micron and other US semiconductor companies building and operating fabs in the US experience cost deltas with our Asian competitors of 35 to 45%. To ensure US global competitiveness, we are calling for an extension and expansion of the expiring semiconductor manufacturing investment tax credit. This will continue to enable the success of America's semiconductor manufacturing renaissance.
▶ 0:37:18Finally, to echo Chairman Gthrey's remarks, having consistent, reliable regulations, particularly in energy and permitting, allows Micron to make long-term manufacturing investments at home so the country can lead in manufacturing and in AI. Thank you, Mr. Chairman. Thank you. Thank you for your testimony, Mr. Tur, Honorable Mr. Turk, you're recognized for your five-minute opening statement. Chairman Guthrie, Ranking Member Palone, and distinguished members, thank you for the opportunity to testify today.
▶ 0:37:46More importantly, let me thank you for this committee's concerted, sustained focus on both the opportunities and the risks surrounding artificial intelligence. As someone who spent a lot of time in windowless rooms, uh, including given my last four years as deputy secretary of the US Department of Energy, let me clearly state my bottom line upfront.
▶ 0:38:06Housing as many AI data centers as possible, especially cutting edge AI training models within our country, is both an economic and a national security imperative. There is no more powerful and transformational technology facing our world. I have also found that the experts who understand AI the best are the ones who most forcefully stress the need for thoughtful effective guard rails and protections.
▶ 0:38:32As the title of this hearing suggests, we need to quickly and affordably convert energy intelligence. The best numbers I found come from Lawrence Berkeley National Lab in terms of what we need to prepare for. In 2023, data centers used 4.4% of the overall electricity in the United States. By just 2028, data cent's total usage will increase to between 6.7 to 12%.
▶ 0:38:59Let me share a three-part strategy to satisfy this increasing electricity demand. First, we need to maintain the full range of tax incentives, grants, loans, and other tools in our tool belt. Now is exactly the wrong time to make it more expensive to bring online new electrons.
▶ 0:39:18Getting rid of just the technology neutral production and investment tax credits 45Y and 48E will substantially raise the costs and delay our ability to power AI. A repeal of just these techneutral tax credits would also increase prices on average US households between $140 to $220 each and every year. Grants and loans including from the bipartisan infrastructure law are also vital.
▶ 0:39:47utility CEOs, developers, rural electric cooperatives are all urging Congress to retain these important tax grant and tools. Let us also remember that among others, the Independent Energy Information Administration predicts that a full 93% of additional capacity added to our grids in 2025 will be with renewables and storage.
▶ 0:40:13uncertainty, whether caused by deliberations in Congress or President Trump's tariff policy, will also chill needed near-term investment to power AI. Second, we need to redouble all our efforts to more quickly permit new power generation and new transmission in our country without sacrificing important Recent bipartisan efforts such as the Baraso Mansion Energy Permitting Reform Act provide a promising foundation for
▶ 0:40:44progress. And third, we should more fully leverage public private partnerships, including with strategic use of federal land for cuttingedge AI, something advanced by both the Biden and the Trump administrations. Ensuring cutting edge AI data centers remain in the United States also gives our democracy a fighting chance to provide effective and efficient guard rails on AI technology.
▶ 0:41:10Companies by themselves simply do not have all the requisite expertise, nor do they have a perspective that takes into account all relevant considerations. We need to fully leverage our biological, chemical, and nuclear government experts to help companies redte team new models to ensure they don't inadvertently empower terrorists and rogue states. We've made some progress, including voluntary cooperation with companies, but we must do more and we must make must also be a top priority to protect public trust and democracy.
▶ 0:41:40Let me conclude by reiterating what I heard from you very clearly, Mr. Chairman, and I think we'll hear again and again throughout this hearing. We are in a global AI race. The stakes are too high for us to lose. I think Dr. Schmidt put it incredibly eloquent with his opening statement. To win, we must all work together, and we cannot take any tools off our tool belt to quickly power AI. Mr. Chairman, Mr.
▶ 0:42:08Ranking Member, and other committee members, thank you again for your diligent, your bipartisan, and your urgent focus on AI. I look forward to your questions. Thank you. I appreciate your testimony. Mr. uh Wang, you're now recognized for five minutes for your opening statement. Chairman Guthrie, Ranking Member Palone, and members of the committee, thank you for the opportunity to be here today to discuss the steps that must be taken to ensure US leadership in AI. My name is Alexander Wang. I'm the founder and CEO of Scale AI.
▶ 0:42:36Today's hearing is personal for me. I grew up in Los Alamos, where my parents were physicists at the National Lab, the birthplace of the atomic bomb. They taught me that America's leadership in science and technology is vital to our national security and global strength. At MIT, I learned that progress in AI depends on three key elements. Data, compute, and algorithms. While most of my classmates pursued expertise in compute and algorithms, few were focused on the data challenge. That inspired me to start scale.
▶ 0:43:07We deliver expert level data and offer technology solutions to leading AI labs, multinational enterprises, and the US government and our allies. At scale, we keep humans at the center of everything we do because AI should always work for us, not the other way around. Over the past decade, it's become clear that the United States faces intense global competition in determining how AI should evolve and who should lead.
▶ 0:43:33In 2018, the Chinese Communist Party's AI master plan started taking shape. They were already developing advanced AI capabilities and using that technology to surveil and suppress their people. Fast forward to today, their plan is more sophisticated and expansive. It includes four key areas of focus. First, the CCP is taking a whole of country approach, having recently launched their AI plus initiative.
▶ 0:43:59Second, the CCP is outinvesting us in data, spending billions on AI ready data and unlocking vast public data sets to fuel AI systems. Third, they're finding ways to catch up on compute and building out their physical infrastructure. And lastly, they're developing leading AI models and exporting them to the world. But we are not here today to just talk about what China is doing, but to identify how the US can lead.
▶ 0:44:24Given how close the competition is across all foundational elements, the policies this Congress promotes could determine the outcome. Global AI dominance is not about trying to level the playing field by mimicking China's authoritarian way of government and AI adoption. Instead, the United States must chart our own course, one that is anchored in American values. This is vital to our long-term national security. This requires decisive action by the United States across four main themes.
▶ 0:44:52Dominate, unleash, innovate and promote. To dominate, we need to win on data. The US government is one of the largest producers of data in the world, but currently most of that data is unavailable to advance American AI leadership. There are three immediate actions that would move us forward towards data dominance. First, establish a national AI data reserve.
▶ 0:45:14This resource should serve as a centralized hub for the government's AI work, housing relevant government data and allowing it to be easily shared between agencies and enabling widespread AI adoption. Second, make all government data AI ready and stand up AI data infrastructure to enable scaled implementation. And third, Congress should invest to position data dominance as a national priority. The next theme is unleash, meaning we must unleash AI technology and establish an agentic government.
▶ 0:45:43An agentic government is one that uses AI under human supervision to enhance its operations. For example, federal agencies could leverage AI to streamline veteran healthcare paperwork, improve improve fraud detection at the IRS, and boost efficiency and information sharing across agencies. This will improve the lives of public employees and the American people. Congress should require each agency to set up at least one flagship agent government program. Next, we must maximize the ability of companies to innovate.
▶ 0:46:13I believe the right regulatory framework is one that allows for innovation while still creating proper guard rails. Congress should take three actions. First, confirm a use case-based regulatory framework and conduct an analysis to address regulatory gaps. Second, establish one single federal AI governance standard to avoid patchwork legislation at the state level. And third, implement policies that enable American workers to become the AI workforce of the future.
▶ 0:46:40These policies would provide the skills necessary to train, fine-tune, and evaluate AI systems. The final theme is promote, meaning we need need to promote US technology globally. Countries around the world, what I call AI geopolitical swing states, will soon be forced to choose between Western or CCP controlled technology.
▶ 0:46:59To help make sure they choose western technology, Congress and the administration should empower NIST to complete all relevant measurement science for AI and export it to the world through the global network of AI institutes. America led the industrial revolution, the space race, the internet and the internet age. AI is the next frontier and with your assistance, I'm confident we will lead again. Thank you for the opportunity to be here today and I look forward to your questions. Thank you. Thank you for your testimony as well.
▶ 0:47:29We will now begin questioning and I'll recognize myself for five minutes uh for questions. So Mr. Dr. Schmidt, you talked about we need all available sources of energy and I think you said in Library of Congress and use AI to side climate change. If we're going to try to build a broad coalition, we can't just go for energy without dealing with climate and we can't just do climate and not have energy because even though wind, solar, and batteries are important, they're they don't they won't supply the energy that we need.
▶ 0:47:57So what do you mean by all sources of energy and how do you think that we could side climate through AI? Um thank you Mr. Chairman. Um as we discussed the de the the needs of our industry are so great that we cannot cut down any of the sources of energy right now. Why does this make sense?
▶ 0:48:19Knowing that climate change is real and knowing that it's a problem is that the intelligence revolution, the ability to do planning and discovery will allow us as Americans to develop new materials, new energy sources and so forth because of the AI data data centers. So our core argument is invest in the way we can now because the future will be so much cleaner and so much more efficient as a result of these algorithms. Thank you for that.
▶ 0:48:48And also um you said in in Library of Congress that Europe has chosen not to grow as we look to our competitor across the Pacific. I mentioned we look over to the Atlantic. What what lessons learned do you think we need to look as we say a lot of times people look at Europe and want to see what they're doing and copy it and what should we not do that Europe has done? Um Europe is a wonderful place to visit but it's not growing. Uh it has great human values but it's not growing. As a result, everyone is unhappy.
▶ 0:49:16The standard of living between the United States and Europe has now diverged. The US lives much better than Europeans which is annoying to my European friends. There are so many reasons why economic growth is important. Growth solves every problem in a democracy. Well, thanks. So you we were talking about that. So what kind of things have Europe decisions European leaders have made that we need to avoid? Well, the primary issue is overregulation.
▶ 0:49:41Um we have a similar problem in America in that uh the overlapping set of local federal state rules which were done with good intentions have the property that they're slowing things down. Uh our competitor China is not a democracy. It's an autocracy whatever you want to call it and they just decide. Um in this fight as I said before if they get there first we will be very upset. All of us will be alive when this occurs. Every one of you will see it.
▶ 0:50:10Imagine a situation where attacks that we cannot even imagine are unleashed by China in an adversarial thing. We have no concept of having a super intelligent opponent where we're not as intelligent as they are. Good thing. So in your article I have here on uh the Atlantic or the foreign affairs, I'm sorry, the foreign affairs. Um you wrote essentially technological advances in the next 5 to 10 years will determine the country gains the upper hand. I have a couple of minutes.
▶ 0:50:38Could you kind of explain militarily what this means and then what because are you right about what they can do militarily and then this is we we need to act now. So um in the framing in China and Taiwan which is discussed a great deal. Everyone assumes that it's a battle of missiles and aircraft carriers. That's not correct. It will be a battle of swarms of drones. Those drones will be highly intelligent, highly planned, and they'll do maneuvers that no one can anticipate.
▶ 0:51:08We collectively are not ready for that. Imagine a situation where China has invented new algorithms for military attacks and defense that we cannot even conceive of. Remember, I'm discussing a world where humans have a partner that is smarter than the collection of those people. As I said, this will happen in our lifetimes, and it's important that we get there first. Uh if you take a look at Ukraine, Russia right now, you see the future of war.
▶ 0:51:32Um I'm assuming, by the way, that China would start by cyber attacks and so forth, there's evidence that uh these new systems will be able to come up with zero day exploits that we cannot foresee. A zero day exploit is something we've never seen before and we can't anticipate. There's lots of people who were worried that that biological attacks can be done and there there's lot there's a a report from the emerging biothreats commission this week with the great details and there's a classified version that all of you should take a look at. There's plenty of evidence that these things are possible.
▶ 0:52:03So Mr. Wang, you see you're shaking your head. I only have about 30 seconds, but if you'd like to make a comment on what he was that comment. Uh I agree with Dr. Schmidt that uh the potential implications of national security are in are incredible. Um as as he mentioned, I think the place we're going to see this first is in cyber. Uh I think we're going to see agentic cyber warfare in which we will see um incredibly powerful AI and large scale data centers being utilized to hack into our systems.
▶ 0:52:29One of the things that we were discussing previously is how vulnerable our energy I'm at zero on my time and we're going to try to stick to it. So I have to hold myself to that. So I appreciate that and we'll get more answers. I'll yield back and recognize the ranking member for five minutes. Thank you, Mr. Chairman. My my questions are of Mr. Turk. You laid out how important it is that we keep our investment environment stable and attractive so AI data infrastructure and energy companies can make the large investments in America needed to build AI tools in the US.
▶ 0:52:58But unfortunately, the Trump administration is doing the opposite. Trump has frozen investments in energy infrastructure, is attacking tax credits for energy generation, and put in place tariffs that are destroying our economy. Don't just take my word for it. Others are saying it, too. So, I'd ask unanimous consent, Mr. Chairman, to insert into the record an article from Politico entitled, "Why Trump's tariffs and tax policies could derail efforts to boost US power supply?" If I can give that without objection. Thank you.
▶ 0:53:28So, Mr. Mr. Terry, can you talk about the harm that some of the Trump administration policies will have on our AI competitiveness? I have to tell you I I' I'd also like to talk about the harm that he's doing to our major universities like Colombia, but that's for another committee, so I can't issue that today. But tell us about the harm that he's doing to our AI competitiveness, if you will. Well, thanks uh uh ranking member Palone.
▶ 0:53:55Uh let me start with the tariffs just because that's the news of the day and the week. Uh I can't think honestly of a worse policy right now if you want to bring on AI power quickly in our country. Power for AI in our country. Uh not only is it increasing costs across the board for the AI and the data centers itself, but for the power that goes into the data centers, but it's also injecting an immense amount of uncertainty. Uh folks who are planning data centers don't want uncertainty.
▶ 0:54:23They want stability of policy so they can plan going forward. So I think tariffs is absolutely the the worst uh if you want to bring on uh additional data and additional energy for data centers. Secondly, the uncertainty of the incentives, the tax incentives, the grants and the loans uh all that Congress has worked on in recent years.
▶ 0:54:43Uh repealing that and even the uncertainty of whether provisions are going to be repealed or not is also going to have a chilling effect on the investment for this power that I think all of our panel here uh agrees that we need to have. We also need to be uh honest with ourselves. Right now, the quickest power, the most affordable power to bring onto our grids, including for data centers, is renewables and storage. 93% uh in 2025 will be renewables and storage. So, we need to focus on a wide variety of energy sources.
▶ 0:55:11Completely agree with folks, but if we want to put urgency to it, the last thing we need to do is repeal these tax credits, grants, uh loans. And then what about I I'm going to ask you to be quick if you will, but because I have a couple questions. What about the repeal of these programs like the inflation reduction act that you mentioned that was put in place by Democrats and the tax credits? How's that going to make energy more expensive for American families in an area of increased energy demand if you would?
▶ 0:55:38Well, we not only have additional energy demand, electricity demand for data centers, uh we have it for additional manufacturing, electrification of buildings. So, the uh demand for electricity is going up now when it's been flat for about 15 20 years. uh that puts upward pressure on prices unless we have more supply coming on. And so to increase the cost of more supply, more of those electrons coming on, makes it more expensive for AI data centers, but it makes it more expensive for households.
▶ 0:56:05I mentioned $220 per household uh on average being increased with just a couple of those tax provisions being revoked. If you get rid of more tax provisions, more grants, more loans, it's just going to increase that cost for everybody. And what lastly, do you agree that we need sensible guard rails to ensure that the privacy and security of Americans personal information is not a casualty of the rap rapid development of these AI algorithms and leaky AI tools, if you will.
▶ 0:56:35Well, I I absolutely agree and certainly uh private companies including those represented here are going to be the ones who do this cutting edge AI and bringing all these tools uh to help humanity. Uh but they have a profit motive. their companies. They are trying to make a profit. They don't have expertise. They don't have nuclear weapons experts. They don't have biological uh weapons experts. Uh we need democracy to step up.
▶ 0:56:57That's why I'm so happy to be in front of you all, a committee that's taking this seriously to have uh hearing after hearing and really uh looking to have that kind of sensible, thoughtful regulation, that balance that the chairman talks about. Well, thank you.
▶ 0:57:10And I have to say again, I don't want to talk about Colombian universities today because it's not in our jurisdiction, but I have to say that, you know, cutting all these research money from major universities trying to get rid of uh you know uh today uh he announced or yesterday that he was denying all the visas for foreign students at Rutgers University in my district.
▶ 0:57:34I mean, I see that we're not going to have, you know, our universities aren't going to be able to do the work that's necessary to actually keep up competitively with China, and it's just really sad. But thank you. I yield back, Mr. Chair. Thank the gentleman yields back and the chair recognizes Mr. Ladder for five minutes for questions. Well, thanks very much, very much, Mr. Chairman. Thanks for witnesses for being here. This is extremely important. And just a few weeks ago, the head of Nvidia said, I'll paraphrase, in order for us to keep the model responsive, we have we now have to compute 10 times faster.
▶ 0:58:04The amount of computation we have to do is 100 times more easily. Another report had come out saying that in 2024 that said that China is looking at about a 94.5% or 94.5% gigawatts new power coming from And so what we're seeing is across everything we've been talking about in this committee for quite a while is that we're going to have to have more power. And Mr. Dr.
▶ 0:58:31Schmid, if I could start with you, you said something very interesting, something that's been brought up in this committee for a good number of years about light touch regulation. And I've heard it from the internet of things that you name what we touch in this committee. But could you just talk a little bit about when you talk about the light touch, what we have to be doing to make sure we stay Um the first thing the government needs to do is to make sure the government understands at the secret and top secret level what China is actually doing.
▶ 0:58:59So some variation of these safety institutes that is at the classified domain that allows our government to understand the details of what our our opponents are doing is important. Uh with respect to the uh current US companies all of them are very well aware of these issues and are working very hard to mitigate them. Um I'm part of a group that actually talks about this informally every week to give you a sense of it. And the companies are trying very hard to keep the models safe.
▶ 0:59:28Having an agreement for example where um the government is aware of what the companies are doing is probably a good thing. That's what I mean by light touch. This innovation, this arrival of this new alien incredible intelligence will be done by the private sector. I want our US government to understand in detail its consequences and help it and help us be successful as a nation. Let me follow up because also you talked about uh you know we need energy in all forms.
▶ 0:59:56A couple weeks ago in our subcommittee on energy we had the RTO's and the ISOs here in this country about seven different ones. They all said this exact same thing. We have to have more power and we shouldn't be taking generation offline. Do you agree with that that statement from all those companies? Well, are all of the ISOs and RTO's? I I don't understand the structure of that part of the industry as well as you do, sir. Uh, from my perspective, the single most important thing to do is to have an all energy strategy.
▶ 1:00:24It as as honorable Turk said, it makes no sense to shut down the renewable stuff. We need more renewables. We also need more natural gas. We need more of everything. We're not going to be able to get the the targets of gigawatts that we need without doing everything more. Right? That includes permitting as I think we've all talked about. Uh thank you Mr. Patia.
▶ 1:00:47You mentioned that we need reliable power and really we can't have dis disruption disruptions out there in it. And uh you know one of the issues again is we've got to make sure again with what the RTO's and the ISOs are all saying is we've got to have this power. Do you see us meeting that power need in the very near future? Because again, when you're talking about, you know, we have to have permitting reform, what's going on in this country? You know, I think that we are um uh we are behind.
▶ 1:01:17I think that we need to think long term, but act now. Um for semiconductor manufacturing, power uh is essential. It's one of the highest input values and it has um incredible impact on the the stability of the power, the reliability of the power has incredible impact on the our ability to run efficient operations and to not have disturbances.
▶ 1:01:41mentioned in my prepared testimony that even a fraction of a second of power droop, not even loss, not even a second, fraction of a second of power droop can have tens of millions to hundreds of millions of dollars of impact in our fabs. So, we absolutely need to make sure that we have more uh more power that that power transmission lines are um you know built for the 21st century.
▶ 1:02:06And in fact, everywhere where we operate our fabs, power reliability is absolutely, you know, at the very top of our list. And we do sight selection. Well, in my last 37 seconds, because you also mentioned we need to cut through that red tape. How would you recommend to this committee that we cut through that red tape? I think one of the ways is making sure that we reduce the duplicative uh uh regulations that are in place between federal and state. Um, in you know, one of our projects in New York right now, we have to do uh similar filings in both the federal and state level.
▶ 1:02:34even though the state level regulatory um uh requirements match the federal ones and so that just you know extends the timeline creates more effort and um you know creates delays and I think the same thing can be true for many many different energy projects and transmission projects around the country. Oh thank you Mr. Chairman my time is expired right on the button and I will uh also submit my questions other questions to the witnesses. Thank you. The gentleman yields back and chair recognizes the gentle lady from Colorado Mr. Deette for five minutes. Thank you so much Mr.
▶ 1:03:03chairman and uh my questions follow very closely on what Mr. La was just talking about. I want to thank you for having this hearing. We had a hearing in the energy subcommittee in 2023 about AI and energy and many people hadn't even been thinking about the the tremendous use of energy by AI up until that time. Um I think that this this issue is a really ripe issue for bipartisan solutions.
▶ 1:03:32However, I'm concerned um sometimes when we talk about cutting red tape or or or um or eliminating um uh these proposals, uh sometimes that's a code word for partisanship. But we have had as as several people have pointed out, we've had bipartisan suggestions. Mr. Peters from this committee has worked on some with people from your side of the aisle. They've had them in the Senate.
▶ 1:04:01So, you know, if if we try to I well, if the Republicans on this committee try to go this alone, then I think this is going to run into trouble. But I think there's tremendous potential for us to work on this in a bipartisan way. Um, I want to talk about an issue though that Mr. Palone raised which is and also um Mr. Turk raised which is is really a concerning issue of today and that and that's these tariffs.
▶ 1:04:26All of the witnesses here today can stipulate that um we're going to need a large increase in all types of energy to not just deal with current demands of consumers but AI. Um does anybody disagree with that? No, no one disagrees with it. So, Mr. Turk, um so if we're going to build more transmission infrastructure, we're going to have to have raw materials like steel and aluminum. Is that correct? Yes.
▶ 1:04:55And um if if these tariffs actually go into effect, which it seems like it might, won't those raw materials needed to add transmission capacity be more expensive? Yes. And uh Mr. uh Batia, just yesterday, in fact, Micron announced that they're going to have price increases on some products today, starting today, doing to pre dur due to President Trump's tariffs.
▶ 1:05:22Isn't that Um we did we um you know memory market is uh is I mean didn't you announce yesterday that Micron is going to impose tariff related search charge on some products from April 9th that the tariffs are an evolving situation and we are communicating with our well Mr.
▶ 1:05:43Chairman, I ask unanimous consent to put into the record a Reuters article which says quote micron to impose tariff related search charge on some products from April 9th. And without objection, so ordered. Thank you. Now um Mr. Mr. um doc Dr. Schmidt, you you just talked about and and I agree with you that um we need more energy in all forms.
▶ 1:06:12Um and and that's and and also that's likely the way that this that this market is going to develop. Is that Yes. Correct. So when people say, "Oh, we need to drill baby drill." That we do need natural gas, but we also need to make sure that we can upgrade our grid to deal with the renewable energy that is inevitably going to be a part of this process. Is that right? Yes. Yes, ma'am. Now, um, so Mr.
▶ 1:06:40Turk, I want to ask you, without guard rails, how how is it that we're going to be able without guardrails that protect consumers? How is it we're going to be able to develop centers, data centers for AI? At the same time, we can ensure average Americans can get the um electricity that they need um at at decent prices. Well, I think we need to do two things at the same time.
▶ 1:07:07uh we need to bring on those electrons as quickly as we can uh including to streamline permitting but to do it in a bipartisan basis. Bipartisan means durable. Uh it means uh making sure things work actually in the real world. And then secondly, we do need to have the guard rails. Um with all due respect to the other witnesses, uh we got phenomenal talent in the US. We're lucky to have that talent working on AI, but we also need to have the government step up. We need to have sensible, thoughtful guard rails to protect everyone's privacy. That's your jobs.
▶ 1:07:37And if we don't have those guardrails, what's going to go what's going to happen for energy prices for consumers? Well, energy prices will go through the roof and we'll lose trust for AI by the American people as well, which isn't going to help our competition with China either. Thank you. I yield back. Gentle lady yields back. The chair recognizes Mr. Hudson for five minutes for questions. Thank you very much, Mr. Chairman. Thank you for holding this very important hearing today.
▶ 1:08:01Uh this topic is crucial for future generations and ensuring the resources uh for healthcare facilities, banks, universities, our national security, including our war fighter. Um thank the witnesses for your very important testimony, very informative. Um bottom line is we must maintain our place as a global leader on AI. I think we can all agree on that. North Carolina, where I live, is a state that leads in innovation and uh that includes in AI.
▶ 1:08:30There's no doubt we'll continue to incorporate AI in many of our industries. Uh but we must ensure we have the resources necessary to advance and sustain AI use. I represent Fort Bragg. We call it the epicenter of the universe only airborne and special forces. At Fort Bragg, we use AI. Um AI benefits the war fighter by anticipating what's next, adjusting to situations, and connecting our soldiers. It's absolutely critical to our national security that we stay ahead of our nearper adversaries, particularly China, so that we maintain our superior advantage.
▶ 1:09:01I never want my guys and gals in a fair fight. I want us to to to be the leader. Um, and certainly don't want us to uh to face a nearpeer adversary that has a superior AI technology. Um, I've also seen threats to our energy sources, whether it's the rolling brownouts we saw in California, uh, but also including when two substations in my district were attacked in my home county, uh, causing a multi-day power loss, uh, nearly 4,000
▶ 1:09:31people without power for almost a week. Hospitals faced blackouts, schools and businesses closed, restaurants and grocery stores lost their inventory. Stop lights were dark. Sell signals went down. Even gas stations had to close. One of my constituents lost her life. Disruptions to our energy supply are dangerous and an attack like this has big implications on our future AI capabilities. The threat only grows as AI is further incorporated in our everyday lives. Um I would ask I'll start with Mr.
▶ 1:09:58Wang, but if anyone else wants to jump in, can you please share uh from your opinion how an unreliable or a non-resilient grid would impact investing investment planning uh and existing commercial activities? Uh first of all, I want to just echo many of your statements. Uh they're spot on. First of all, we need advanced AI for national security. We need our Department of Defense, our our war fighter to have advanced AI capabilities. Um that is absolutely critical for this next phase.
▶ 1:10:27And uh and that is dependent on energy as as we've discussed here in the uh uh over the course of today. Um, one of the one of the greatest risks if you think about the training of these large-scale AI systems, it requires a continuous source of power to be able to uh both train advanced AI systems and keep them running.
▶ 1:10:45Um if we have an unreliable energy grid in any sort of uh you know uh competitive or conflict scenario um if the adversaries have the ability to take down our grid through cyber attacks or other forms of of attacks then that greatly impacts our ability to be competitive or or to be able to fight uh fight in that battle. So um it is absolutely critical we have reliable energy grid. It's important that we secure this energy grid.
▶ 1:11:12It's important that we're able to protect against cyber attacks and other uh other forms of attacks and we have consistent power. Does anybody else want to weigh in? I think that you know the um uh the president as well as uh Congress is um behind uh the strong growth in manufacturing across many different sectors including semiconductors and energy uh has always been an advantage for this country um due to you know abundant natural resources and we've invested
▶ 1:11:43in in all different technologies uh over time and that has just stalled over the last um 101 15 years as some of the other panelists noted and So absolutely believe that investing in energy is going to um help to to fuel this manufacturing renaissance and in particular the semiconductor renaissance that um you know we all know is so important to to winning in AI. AI is you know just to to um co-op some of the words of um uh Mr. Wang on the panel here.
▶ 1:12:10You know, it's about data and data needs memory chips and it needs um the most advanced memory chips in order to be able to um create all of the insights that u are are valuable in whatever circumstance or application that we see. Well, thank you for your answers, Mr. Just Congressman, if I could just say a word on grid. And I'm just so pleased you mentioned the grid. It's just such a fundamental backbone of our our infrastructure for military bases, but for everybody else in industry.
▶ 1:12:38Uh this is where I think it's so important to have all the tools in the tool belt. Uh you all provided through the bipartisan infrastructure legislation $10.5 billion dollars to improve our grid through a program called the GRIP program. Uh we have $23 billion in our loan program right now with a whole bunch more in the pipeline to help utilities to strengthen the grid going forward. That's what I'm talking about of keeping all these tools in the tool belt. Thanks. Appreciate that. Mr. Chairman, my time is expired, so I will yield back. But thank you to the witnesses for those answers. Thank you. Thank you.
▶ 1:13:07Uh the chair will gentleman Leilsback chair recognizes Miss Shakowski for five minutes for Thank you s so much. Um Mr. uh Perk I want Did I do that wrong? No. Okay. Um I have questions for you. Some go way back.
▶ 1:13:26We're talking about AI today, but I have to tell you that I and Gus Billis on the uh Republican side have been working on the issue of uh of privacy for a very very long time and even be more before that with uh in all kinds of tech interests. But we have never done anything to reign in big tech.
▶ 1:13:55Nothing whatsoever. So we see families um that have to give all kinds of information which they they do all kinds of I think risks that that go on. So, of course, we're we're talking about AI, but in the meantime, we have seen tech leaders um apologize to consumers.
▶ 1:14:23Oh, we didn't mean to put children at risk. We didn't mean to do this or or or or that. And yet, we have done nothing at all. So what it seems to me is now we're talking about AI and you talk about risk. You use that word risk. So do we have to go back further or let's we if you want to just talk about risk with with AI.
▶ 1:14:53What exactly are are we talking about? You you mentioned that in your written statement. Um, but I'd like to know who, and we can start with how do we address the issue of risk, but who should be addressing the issue of risk? Well, thank you, Congresswoman, for that question and thank you for your focus on these issues for years uh and years.
▶ 1:15:22Uh your leadership has just been uh tremendous. Uh I completely agree with something Dr. Schmidt said at the beginning in his opening statement about uh AI being underhyped if anything. This is an incredibly powerful technology. What that means to me is and I've had the chance to work with a lot of our experts in the government and we need to make sure that we keep those experts in the government.
▶ 1:15:43We need more AI experts in the government not letting uh AI experts leave which gives me concern uh with some of the firings and some of the other things that this administration the Trump administration's doing. Um, but powerful technologies can not only be used for good. Powerful technologies, especially in terrorist hands and rogue states hands, in other hands, once you get these models out there, uh, it's incredibly difficult to bring them back in. So, I'll give you a specific example of a risk.
▶ 1:16:10And I know this is an unclassified setting, so I'll just talk a bit in generalities. uh as smart as the folks are who work in scale AI and open AI and Google and Meta and uh these other big tech companies, uh they're not nuclear weapons experts, nor should they be. I'm not sure why you'd be a nuclear weapons expert. And uh Mr. Wang mentioned his uh parents working at Los Alamos, which is just a phenomenal lab uh for us in our country.
▶ 1:16:35We need to make sure that before a model goes out there in the public that there's some red teaming, there's some vetting by nuclear weapons experts to know what to look for to make sure that terrorists can't take these models and help them develop nuclear weapons or biological weapons or chemical weapons. That's where I think they're just as you suggest it's it's who and how uh the private sector will need to leave. They have an incentive to make sure that their uh models are safe, but they don't have all the expertise. They need to red team and make sure that those models are safe.
▶ 1:17:04uh I would prefer that not to just be a voluntary uh kind of understanding. I think it should be a requirement that before models come out there needs to be some vetting. Now that has to be done efficiently. It has to have the right balance that the chairman's talking about to make sure that the broad expertise that we have the nuclear weapons experts the chem weapons experts are uh poking and proddding and making sure that these models aren't going to cause us uh harm. So that's one particular example.
▶ 1:17:29The other one that you mentioned which I think is incredibly important as well is uh privacy and making sure that information is not sucked up inappropriately to go into these AI models in the first place and that these models aren't going to infringe on the privacy. I absolutely think consumers, citizens, Americans should control their information and we should have guard rails. We should have regulations in place to ensure that that's uh that's the case. Again, this should be thoughtful. This should be efficient. This should be in a way that allows our companies to push the boundaries.
▶ 1:18:00I completely agree with everyone who's spoken that we need to win this AI race, but we need to do this thoughtfully and make sure that the democracy uh the people's representatives have some say here, too. So, as part of the hu the who you're saying that the Congress of the United States should play a role. Absolutely. That's why I'm so pleased that this committee is uh having multiple hearings, not just one off. Thank you. Well, thank you so much. I appreciate it. I yield back. Thank you. The gentle lady yields back. The chair recognizes Mr.
▶ 1:18:30Bill Ruckus for five minutes. Thank you, Mr. Chairman. I appreciate it. Thanks for holding the hearing. I thank the uh the presenters. Uh Mr. uh Wang, to win the race against China, American AI companies need to succeed at home and abroad. However, we've seen our largest foreign trade partners, especially the EU, enact sweeping new AI data regulations that could be used to target US companies.
▶ 1:18:58How can we address new and emerging digital trade barriers to ensure American AI companies can out compete their Chinese competitors uh again on an even playing field? Uh thank you for the question. Um you know it is certainly true that uh that China in particular and the Chinese Communist Party has a strategy to win on data.
▶ 1:19:24Um this includes some of the things that you mentioned around uh you know being more loose around data privacy both domestically and internationally as well as explicit programs that they have within their country to create tax incentives um you know vouchers and other forms of of large scale government programs to to win on data. Um that's why I actually think it's critical in the United States that we focus on exporting our technology globally as well as exporting our standards globally.
▶ 1:19:51So uh one of the one of the avenues that we have as a country to be able to do this is through NIST. um you know as uh as the United States we have an opportunity to really define what are the standards for AI that will be adopted globally and other countries are listening and they you know through the global network of AI safety institutes there is a a global coalition of countries who are looking to us to help define what are the standards for safety and other provisions that they'll that we will all collectively
▶ 1:20:22utilize to define how we govern AI in the future. So we have a golden opportunity as a country um to help set the global AI standards um and we need to take that and be very thoughtful about what we what we present. Thank you. Uh Mr. Schmidt uh Dr. Schmidt uh we've seen Chinese AI companies deepseek uh Alibaba and now Manus I AI uh debut powerful AI models in the past two months.
▶ 1:20:51Many are rightfully focused on these models capabilities, but I'm also concerned about how they were trained potentially on Americans personal data and by misusing access to American AI services. We should also be cautious about how Chinese AI will be used by American consumers and in potentially sensitive areas of the US economy such as healthc care. And I'm very interested in that.
▶ 1:21:20We should act now before China has a foothold on these emerging markets and controls AI data outputs to Americans queries. Uh what steps can we take to address these risks to American consumers and businesses? And first I want to talk to Dr. uh if Dr.
▶ 1:21:44Schmidt can answer and then if anyone else wants to chime in and I have some time uh please don't hesitate. Not only is your question great, it's worse than you said. Sorry. The Chinese models are released in open source which means that you can see how they work and they're easily spread and they're free. It's highly likely that the US companies will be by the time we're done pretty well regulated by you all for because of the importance of what they're doing.
▶ 1:22:13This is my personal opinion. I'm not calling for it, but I think that's what's going to happen. It's very hard to regulate the open source movement um coming out of China. We need to make sure we deal with that. the industry is struggling with your question because we've not figured out a way to deal with what is called distillation where distillation is where you take one model and you ask it questions and you get the answers and the ch there's lots of evidence the Chinese did exactly what you said in your question using this distillation meth mechanism so we don't really know my
▶ 1:22:44own view is that the best answer is more offense not more defense and simply invest invest invest to stay ahead in order to invest as I mentioned in my opening statement. We need high skills immigration of key people because these things are essentially math problems. We need all the energy that we discussed. I think the American innovation system which is the combination of the government, the venture capital industry, the private sector, universities is phenomenal. It's important we not in any way slow down the universities and AI research.
▶ 1:23:14We can win this. It's has to be an allcountry effort. I my personal view is it's a national security issue for America against China. Agreed. Agreed. Yeah. I don't have any more time left, so I'll yield back. Thank you, Mr. Chairman. Thank you. The gentleman yields back. The chair recognizes Miss Matsui for five minutes for questions. Thank you very much, Mr. Chairman. I want to thank all the witnesses for being here today. U this is a critically important issue for the future of America and really for the future of humanity at large.
▶ 1:23:44Now, as co-author of the original Chips Act, I know how critical this policy is to strengthen our national security and technological leadership. The Chips and Science Act is working as intended, leveraging it $50 billion of government funding to spur a ninefold investment of $450 billion from the semiconductor industry.
▶ 1:24:05Yet, President Trump threatened to abandon this once in a generation effort to to bring advanced semiconductor manufacturing back to America. And his tariffs are driving up cost to what we need to be competitive in AI, including aluminum, steel, semiconductors, and electronics. This administration's chaos and uncertainty will harm our AI leadership. Dr. Schmidt and Dr.
▶ 1:24:31Batia, how would dismantling or delaying the Chips and Science Act programs impact America's global competitiveness, especially in AI Thank you, Congressman Matsoule, for your support of the U. the Chips Act and our um and our industry uh over your um over your entire career.
▶ 1:24:51Um we are the only US memory company and our investments um are uh you know bring tremendous value across leading edge memory solutions as well as across other industry other segments like the automotive industry, aerospace, defense.
▶ 1:25:05So we we believe our investments and our more than hundred billion dollar plan over the next 20 uh years will bring tremendous value and we're actually encouraged by the um the executive order to uh create an accelerator program for largecale projects through the chip's office through the department of commerce to be able to um ensure that our projects so you don't want any slowing down right no pausing. That's right. That's right.
▶ 1:25:33I think the accelerator's goal is to be able to make sure that the projects can be successful. Dr. Schmidt, um, a number of us were very strong supporters of your legislation for the following reason. Uh, 25 years ago, we made a mistake as a country, and we got out of this business. It costs money to get back into it. It costs money to build the factories, to train the people, and so forth.
▶ 1:25:5610 billion of the 50 billion is in new R&D for new kinds of packaging which will give America a possibility of leading globally in semiconductors. The other 40 is to allow us to have domestic production. Why do we need domestic production? Think national security. Just think about it. It's worth it to our nation to have a supply chain of critical intelligence materials. That is literally the things that do the thinking under your control, Congressman. Okay. And um Dr.
▶ 1:26:25Schmidt, I read your testimony. I was very impressed with it, particularly in a part about um our innovation power, the potent collaboration between government, private industry and academia. Um the government, you know, really providing the strategic direction and uh the private sector driving innovation and academia which fills a pipeline of foundational research and talent.
▶ 1:26:52I was wondering um you know I think it's great to have this collaboration but I'm wondering because the Trump administration claims are committed to American dominance. Yet time and again their actions show the opposite. We should be training and recruiting talent to shape our AI leadership. Instead more than 75% of US researchers surveyed are considering leaving our country because of the chaos of the Trump administration.
▶ 1:27:19President Trump is firing experts in our agency, waging a war against science and destroying our public research funding system. Mr. Turk, this dismantling of public research and reductions in the federal workforce consistent with is it consistent with strengthening US leadership on AI and other emerging I think it's absolutely inconsistent. Um, I think this is a huge threat going on right now. Hollowing out the expertise uh in our national government.
▶ 1:27:49I certainly got a chance to work with phenomenal experts, civil servants in our department of energy. Uh, we built that institution up for years and years that kind of talent at our national labs and to be haphazardly and chaotically uh firing people, losing that talent at the exact time that we need it given the global competition we've got in AI and all these other critical uh technology areas. So, I think it's exactly the wrong approach. Okay.
▶ 1:28:15Um, there are other energy technologies the Republicans have historically supported. The bipartisan infrastructure law, we created the office of clean energy demonstrations to help develop advanced nuclear, hydrogen, carbon capture, and longduration energy storage. Mr. Turk, what's happening to the office of clean energy demonstrations under President Trump? So, it is uh one of the offices that's been decimated uh the worst.
▶ 1:28:42Uh and you just mentioned the incredibly important areas that they're working on. Funding that's been provided on a bipartisan basis from Congress and to see that paneling is uh is just a move on. My time has I have other questions I'll submit for the Thank you. Thank the gentle lady yields back. The chair recognizes Mr. Palmer for five minutes for questions. Thank you, Mr. Chairman. Um Dr. Smith, I read your book and um I have one of my favorite quotes is from Henry Kissinger. You said the absence of alternatives clears the mind marvelously.
▶ 1:29:11I I say it another way, nothing brings clarity and focus quite so well as the absence of alternatives. Um my concern is that we're in an arms race with China for artificial intelligence and quantum computing and that if China wins that race, they will not be a superpower, they will be the superpower. I also continue to point out in this committee that uh there's not a single major refinery for rare earth elements in the western hemisphere. There's only nine in the world. Eight are in China, the other one's in Malaysia.
▶ 1:29:40And um I just want to ask you, do you think this ought to be one of those moments of clarity that uh focuses Congress on on meeting these these demands, these needs? Um thank you. And I do if I told you with certainty that in 5 years China will be able to mount cyber attacks against American in infrastructure that we have no defense of, would you act now? Yes, absolutely.
▶ 1:30:10If I told you that China was building an architecture for national security that was autonomous, robotic, uh, attitudable, etc. Would you act now? Yes, you would. I'm telling you those now. So if if we don't act on uh the mining, processing, refining of rare earth elements immediately, we could find ourselves in the very position you just described. Um that's correct. We want full control of our own supply chain. Absolutely.
▶ 1:30:40Energy, chips, the infrastructure that we need. It's an issue of national security for America. Mr. wing in order to meet the uh demands that that we have for power generation. How what power generation capacity do we need to have to to achieve dominance in AI and quantum computing? Do you have do you have any idea of what that would be?
▶ 1:31:02Uh well, as was mentioned uh earlier, the the scale of data centers that are being built require similar amounts of power as entire cities um in the United States. Well, Dr. Schmidt. Uh I don't you probably don't remember this at the dinner at the Library of Congress. You and I had a brief discussion. One of the things that I continue to point out in this committee and other places is that there's 100200 hydrocarbon uh power generation facilities that have been dismantled.
▶ 1:31:29U we know that uh we we have these enormous power demands. I know there's a move now to go back to opening these back up on natural gas and coal, but what do you think about um using small modular reactors to locate them on these facilities to meet it's the quickest way I think to meet these these power demands and and the good part of this is with all due respect to my Democratic colleagues, we're not going to do it with renewables.
▶ 1:31:57Uh because we just don't have the time to build out everything you have to build out, including the transmission lines. Those transmission lines still exist at these shuttered power plants. We could literally uh we could open them with coal or natural gas, but I think we ought to be thinking about small modular reactors that can plug into to the existing transmission lines. How would you respond to that? One of my personal frustrations is the regulatory structure around nuclear and SMRs.
▶ 1:32:26Um SMRs are the right answer. So your your instincts are exactly correct. Um, furthermore, they can be built in volume. How many SMRs are in use in America today? Zero. Zero. How many where what is the most promising one? An initiative in Canada. Um, I'm glad you brought that up because they just licensed it what, two days ago or a week ago. And the typical supply, the fast approval time is considered to be 12 years. That defies logic.
▶ 1:32:55We need a new program around much faster permitting for safer and safer fision and fusion nuclear. SMRs are the correct path. Uh one of the issues that's sorry for the details is 30 years ago, 40 years ago when there's the standard for permitting in nuclear was set at a threshold below natural radiation. Alex can talk about this with great uh detail more than I can. At the end of the day, it was set too hard. It was a mistake. Needs to be fixed.
▶ 1:33:22Well, uh, the GE Hitachi and I'm taking sides for any brand could be built in about three years. But you made another point there that I think is very important for this committee and that is the economy of scale. If we were committed to building these out in scale uh uh so much of it can be done in factories, so much of the testing can be done in factory and then on site.
▶ 1:33:43I think it is extremely important that that this government move towards small modular reactors to meet the power demands that we have to have to even be competitive with China and in the AI space. Thank you, Mr. Chairman. I yield back. Gentleman yields back. The chair recognizes Miss Caster for five minutes for questions. Well, thank you, Mr. Chairman. This is a very important topic and we should be focused on bipartisan solutions to advance American innovation.
▶ 1:34:10The problem is there are so many new roadblocks right now and President Trump has turned himself into the anti- innovation president. Uh he is has outright killed large new energy resources that were in line to come onto the grid. Uh he's imposed these new import taxes and tariffs on everything we need to compete on AI, aluminum, steel, semiconductors, electronics. uh he's threatening to halt our investment in semiconductors in America.
▶ 1:34:41Uh he is u has taken a hatchet to the academic and scientific workforce. This is all a gift to China at exactly the wrong time. But let's focus in on the the challenges and the opportunities for energy and AI. Um Secretary Turk, it's good to see you. Uh, one of the challenges is, uh, the enormous need for new energy capacity, but I'm very concerned for what this means for everyday Americans and their electric bills.
▶ 1:35:11There, Mr. Chairman, I'm going to offer ask unanimous consent to submit for the record a new study from the Environmental and Energy Law Program out of Harvard Law School. um where they highlight they say they're skeptical of utility claims that data center energy costs are isolated from other consumers bills rate structures as well as secret contracts could be transferring big tech's energy costs to the public how how do we balance objection order thank you how do we
▶ 1:35:41balance these these needs so I I think we have to um as I said in my opening statement uh bring power uh on the grid uh could be behind the meter as well to power data centers to pay to power AI cutting edge uh models. We also need to make sure we have downward pressure on prices.
▶ 1:36:04I don't know of any elected official anywhere in our country who shouldn't have affordability and the cost to consumers as job one and everything seen through that lens. Uh as you suggest, it doesn't seem like that's what the president our president right now has in mind. So even contemplating repealing the tax credits that uh puts downward pressure on prices across the board, technology neutral, right?
▶ 1:36:31Any technology can qualify for those tax uh credits if it meets certain thresholds. Um getting rid of those is exactly the wrong thing to do right now. And I mentioned uh the analysis that's been done. A number of groups have done really good cutting edge analysis. uh 20 $220 more annually each and every year for an average household. Now, that goes up in some states to uh $400 uh more a year.
▶ 1:36:57Uh if uh you happen to represent Missouri, Arkansas, Texas, New York, Iowa, and Kansas, watch out. It's $400 more per year just to repeal two of the tax credits, let alone the full paniply of what's uh what's been done. Yeah. Thanks. So, one of the opportunities, however, is to work together on a much more uh efficient and modern electric grid across the country. It's kind of outdated the way everything is structured and right now.
▶ 1:37:26Um, that's why yesterday I introduced my advancing grid enhancing technologies act with senators Welch and King that will implement shared savings incentives that promote the deployment of grid enhancing technologies. That is the cheapest way to supercharge our grid. We've got to optim optimize the existing grid infrastructure to bring energy projects online more quickly and save consumers billions of dollars.
▶ 1:37:55Do you see uh hope here with our with modernization of the grid and and gets well thank you for your leadership and thank you for focusing on gets grid enhancing technologies. Um we've got such a range of technologies. Some we still needed to reduce cost but some like gets and reconductoring makes sense. Uh we just don't have a utility industry now in the incentives for those technologies to be utilized at scale to allow us to get more out of our existing grid.
▶ 1:38:22We of course need to build new transmission as well in our country to make sure that we're prepared for what we need in the future. So I'm uh really pleased and uh thankful for your leadership in that area. I'm also firmly one and I agree with Dr. Schmidt and others uh the Congressman uh Palmer was just talking about small modular reactors. Uh I think it's small modular reactors. I think we should be investing now in fusion so we have that as a solution.
▶ 1:38:44Enhanced geothermal is such a phenomenal resource in our country taking advantage of the drilling expertise in the workers uh to drill 247 clean power including for data centers. Uh we should be investing even more now to try to bring that technology online very quickly. Thank you. I yield back my time. Thank you. The gentle lady yields back. The chair recognizes Dr. Dunn from Florida for five minutes. Thank you very much, Mr. Chairman.
▶ 1:39:07So exciting topic from tech startups to energy uh grid operators, internet service providers, everybody is uh working to develop these new AI technologies and America has always been at the forefront of technological innovation. But with AI, we're just not we're not untouchable. We uh we we have competition. We're in a race with China to lead in this field. And it's promising to have two major American companies sitting before us today.
▶ 1:39:35scale AI and micro uh who are leading the way. This global AI boom has prompted widespread industry adoption across all kinds of sectors. Healthc care is one of keen interest to me but also finance, telecom, weather. This morning I met with Noah. They're excited. Um however this exponential growth of demand brings it with some substantial energy requirements and as AI models grow in size and complexity so does the infrastructure required to train an operator.
▶ 1:40:06For instance, training large language models can take weeks of processing and high powered GPUs and the energy consumption can be staggering. At the same time, our telecom's infrastructure has to keep up with AI's growing demands. High capacity networks are essential to ensure fast data transfers and these real-time AI applications such as autonomous driving, tele medicine and smart cities and whatnot.
▶ 1:40:32As AI use grows, both the energy consumption telecoms capacity required will grow commensurately with it. At the same time, the Chinese Communist Party is moving fast and hard with zero regulations and zero ethical restraints. Uh so we have our work cut out for us.
▶ 1:40:51I also sit on the China select China committee and the NATO parliament and I had a chance to discuss these issues with our European counterparts and I met with the member of the European Parliament who uh led the current uh EU privacy bill and she cautioned me whatever you do in Congress don't do that. Don't do what we did. Don't do what we did. That was her words. Uh kind of from a right from the horse's mouth if you will.
▶ 1:41:18The uh the expert witnesses here I think understand today that the EU bill has indeed restricted artificial intelligence development in Europe. With that, Mr. Schmidt, as these AI tools develop, their utility to each of us will be proportionate to their our ability to access them. With that in mind, are we moving quickly enough to uh enable the deployment of broadband connectivity uh and commercial access to spectrum? Mr.
▶ 1:41:48Um on the spectrum side um we need another round of a spectrum analysis and a new way in which the unused spectrum is allocated. I happen to believe in a situation where companies are able to uh buy the spectrum but they have to build it out or they're given the spectrum they have to build it out. I don't want people sitting on spectrum and not making it use. We need that bandwidth. However you all arrange that it will be fine. fine with us, I think.
▶ 1:42:18Well, it's trickier than you think, but thank you for that. I'm pleased that this where our members of that committee are sitting here with us today. Uh Mr. Wang, it's good to see you again here in Washington. You've becoming a regular up here. I fear for your soul. Today, cutting edge AI research is dominated by industry partially due to the very high cost of computing needed to train these advanced models.
▶ 1:42:42uh given the fast pace of the progress, how can we ensure our government or our pilot programs keep up with the rapidly evolving industry needs and standards? I think the the most useful framework here is to just think about what are the raw ingredients uh for these AI models. So it boils down to three major elements. Uh computational power which requires a lot of energy as we've discussed a lot today.
▶ 1:43:07um algorithms, the sort of instructions for the models uh and that requires incredible talent to devise new algorithms and then data and uh oftentimes we really under and and really these AI models and progress in AI models boils down to progress in every one of these three underlying uh components. Um oftentimes we don't consider enough our relative position on data with respect to the Chinese Communist Party.
▶ 1:43:33Um they have had a decadel long strategy to be dominant in data to win on data. Uh they have large-scale government programs. Um they've built their entire system and their entire country their sort of civil mil military fusion system to be uh dominant on data and we need to begin thinking as strategically on this front as well. We need a program and we need thought around how we achieve data dominance as a country.
▶ 1:43:58how we utilize all of the incredible data that we have as a country to to get out ahead. Um, and our government is one of the largest producers of data and we need to leverage this advantage. And do you think having a privacy law would help that? I mean, a standardized privacy rule for the country. Uh, I definitely want to prevent this the case where we have a patchwork of privacy laws across every state in Jim's out of time. I'm I'm sorry. Okay, just on that sort of on Jim's out similar question.
▶ 1:44:27Coincidentally, out of time. I'm sorry. I'm sorry. Announced yesterday. I'm sorry. You're out of time. Am I out of time? God, that went flew fast. Yields by fast, as it Mr. for five minutes. Thank you, Mr. Chair.
▶ 1:44:40Um, let me begin by acknowledging that just about every witness who has testified at the energy subcommittee this year, whether by invitation of Republican or Democrat, has agreed that we must make it easier to build transmission infrastructure to meet our nation's growing energy demands and to be the global leader in AI. Today's witnesses are no exception. So, I really hope this is an issue that the committee can get serious about. Uh, Dr.
▶ 1:45:06Schmidt, your testimony mentioned building more transmission, but you also called out the need to embrace small grid capabilities and grid enhancing technologies. Why is it important to maximize our existing electricity systems efficiency and performance while we also work to build new One of the ways to think about the energy problem is you're building things that last 40 years and that you're in a constant process of renewing things that were built 40 years ago.
▶ 1:45:34Um, and in that sense, we need an integrated plan to upgrade everything. I I like what what uh the honorable Turk said that you need all of it. Um, my list was fusion, fision, especially SMRs, enhanced geothermal, natural gas, renewable wind, and solar. Uh, we need all of it. In order to do that, the grid has to be more dynamic. You want to have the source of of power be as close as possible to the consumer. The ideal scenario is you put your power plant next to your data center.
▶ 1:46:04The data centers need 5 gawatts. They're huge, right? You need 5 gawatts of power, which is also huge. We can't do that. Therefore, we need to have the transmission to get them from one to the other. Thank you. I appreciate that. And Mr. Turk, your thoughts on this? If we can make some existing loads more flexible through demand response programs or deploy gridenhancing technologies to get more out of our existing infrastructure, are these important tools to create the energy system conditions needed to win the race for global AI leadership?
▶ 1:46:34Well, I think they're absolutely indispensable and they're no regrets as well. Just as Dr. Schmidt was talking about, why wouldn't we take advantage and I know you've been a leader on this for years in the Congress. Why wouldn't we take advantage of that infrastructure that we built? Uh infrastructure is one of those things that's going to be around 40 years or even more. Let's get the most out of it. Uh and we do have technologies these days. Uh gets technologies reconductoring using AI, using machine learning to help the grids uh balance loads a lot uh a lot quicker.
▶ 1:47:03uh we started a program at the Department of Energy to use AI for permitting to make sure that we could do more permitting including on transmission to build out our transmission system even more uh more quickly uh than we uh than we have been doing. It's a big challenge but uh we got to use all the tools. Thank you both of you gentlemen. And Mr. Batia, I want to first and foremost welcome you to upstate New York. Um I also want to thank you and recognize the important commitments that Micron has made to upstate New York.
▶ 1:47:30Today we're talking a lot about how we can build and operate our energy system to meet AI's needs. But we rarely talk about how AI and its enabling technologies can be developed to better fit within the reasonably foreseeable constraints of our system. And that is why it is critically important that we continue to invest in research. So Mr. Batia, your testimony mentioned that Micron is developing chips with much more um improved energy efficiency.
▶ 1:47:57Can you discuss why this is a priority for Micron and how important is a chip's energy efficiency to reduce the overall energy demands of these data Certainly thank you uh for your uh comments congressman and uh you know I I really believe the semiconductor industry and memory chips are part of the solution.
▶ 1:48:16Um you know the brilliance of Moor's law uh which is the the governing law for the industry over the la since its you know inception 50 plus years ago is that with every generation of technology that we introduce 18 every 18 to 24 months we are taking the same operation and doing it with higher performance lower power and less you resources utilized to build that that device whether in our case memory cells and so that scaling path by itself is part of the
▶ 1:48:47solution to being able to to make all these tremendous AI innovations, these datadriven AI innovations come to life using lower and lower power as we progress through time. And Micron has actually um been very focused on leadership in that uh in that way. For the last four DRAM generations, Micron has been first to market by several quarters ahead of our Asian competitors. And that allows us to build chips that are lower powered than those competitors.
▶ 1:49:15So for example, I mentioned in my prepared remarks that every one Nvidia GPU has 96 high bandwidth memory chips integrated with it. Our high bandwidth memory chips are 30% lower power 30% lower power than our competitor competitors chips that go into those similar systems.
▶ 1:49:36So absolutely critical for us and we we look for all avenues to be able to to continue to reduce power as we scale down um the trajectory and and improve the efficiency of our of our um ships including partnering with national labs do research with um time has expired. Thank you. Well thank you very much Mr. Chair. Thank you for the comments from the gentleman yield back. Thank you and the chair recognizes Dr. Joyce for five minutes.
▶ 1:50:00First, I want to thank Chairman Guthrie for holding this critical hearing on the future of artificial intelligence. AI is the defining technology of the next several decades. It will have a revolutionary effect on all aspects of our lives. It will be integral in everything from the high level data analysis to the use of a search engine.
▶ 1:50:23In industries as diverse as energy production and health care, AI is already making significant inroads. As a doctor, we see AI integrated into innovative medical devices, helping to translate the information collected by the device into clinical guidance. In medical practices, AI can help streamline the administrative tasks, allowing doctors to ultimately spend more time with their patients.
▶ 1:50:52This is just the beginning of the capabilities that AI will give us and it is why it is critical that the US leads the way in the development and the deployment of this technology just like the space race during the cold war. However, our geopolitical rival is striving to catch up and overtake America so that they can dominate this new sector. Make no mistake, China is desperate to beat us in the field of AI.
▶ 1:51:19It is a national imperative that we do not allow this to happen. America and the free world can't afford to let the Chinese Communist Party win the race with AI. Fortunately, we have an advantage. And that advantage is the vast energy resources, the resources that are under the feet of my constituents in Pennsylvania. Energy is now the limiting factor for building the data centers that AI uses.
▶ 1:51:47Which is why to win the race for AI, we need to unleash American energy. We've already begun to see the new project development with data center agreements between AWS and Talon Energy at the Suscuana Nuclear Generation Facility and the reopening of Three-Mile Island thanks to the power purchase agreement between Constellation Energy and Microsoft.
▶ 1:52:12Another project in Indiana, Pennsylvania was the announcement to repurpose the retired coal powered Homer City Generating Station. This new facility powered by Pennsylvania's abundant natural gas reserves will be one of the largest power generating sites in the entire country, capable of generating up to a staggering 4 and a half gawatts of electricity to power data centers and AF AI facilities on the site, attracting billions
▶ 1:52:42of dollars in investment to our region. These projects show that America's ability to lead the world in AI is directly tied to our nation's energy production. We must continue to use our energy advantage in this global competition. Dr. Schmidt, in earlier public statements, you had supported moving away from fossil fuel base load power. Today, it seems that you have a different view on the energy industry.
▶ 1:53:09Can you please explain why your views have evolved and what that connects with your views on AI development? Um, let me also mention that 35 to 40 years ago, Carnegie Melon in Pittsburgh invented a great deal of the world that we're talking. So, thank you to your state and to what they were able to do. Um, we need all sources of power to accelerate because we don't have a choice.
▶ 1:53:36uh if you just assume that you can get there with base load power with renewables you can get there maybe 25 30% we can debate it you can't solve the whole power the whole problem as we're laying out without an all power solution and that's why I'm taking the position that I have today you mentioned renewables but renewables do not provide that base load power that is so necessary in the data centers correct uh not correct I'm sorry sir um renewables plus batteries are now roughly competitive with the price of new um
▶ 1:54:07uh new u natural gas partly because the natural gas demand is gone so much. And so again, from my perspective, the answer is yes to all. Let the market sort it out. Let everybody build everything. We need it all now. And I think ultimately we need the nuclear and the natural gas to be able to allow those data centers to continue to develop and continue to grow. Yeah. And let me sir may may I just emphasize the importance of base load power which I think is what you're getting at.
▶ 1:54:35We need continuous if you listen to Micron these guys are are superhuman. What they've done in America against the Chinese and the other Asian manufacturers is enormous. They need that base load power. That's why your premise is correct. Thank you very much. I think we can all agree that the base load power is truly the key to moving forward with the development of AI in the United States. Mr. Wang, as I mentioned earlier, there are two data center projects in Pennsylvania that are coll-locating with nuclear power station.
▶ 1:55:04I'm sorry, we're we're beyond time. Sorry. Thank you. I will my questions for followup. I thank you again, Chairman Guthrie, for holding this important hearing today. Thank you. The vice chair of the committee appreciate yields back and the chair recognizes Miss Kelly for five minutes for questions. Thank you, Mr. Chair. As my colleagues have noted, this hearing comes at a pivotal time.
▶ 1:55:27The Trump administration, led by Elon Musk and Doge, are working to resend key investments made under the Inflation Reduction Act, which you've heard, that have spurred unprecedented growth in clean energy generation while expanding domestic manufacturing opportunities. These investments have made a major difference in my district, which is urban, suburban, and rural. I go from the city of Chicago downstate where I have 4,500 farms. Mr.
▶ 1:55:55Turk, given the expected growth in demand and significant investments that would be made to our grids infrastructure, can you discuss the difference the different responsibilities between states and federal governments in regulating how these improvements will be paid for? Yeah, absolutely.
▶ 1:56:13Uh luckily we've got policies in place, tax incentives in place, grants in place, loans in place uh to make it more affordable to uh uh for us in our country to uh build the kind of power that we need, not just for data centers, but for the rest of our economy as well. Repealing those tax credits, I've hope I've been incredibly clear here at this hearing.
▶ 1:56:36repealing those tax credits, those grants, those loans, uh, will raise the price, will raise the costs, and will delay how quickly we can bring electrons, uh, onto our grid. So, I think it's incredibly important for the federal government to play a strong role. Luckily, we've got those incentives in place. It's just a question of whether we uh take those off the table, take those tools off the table. And I just couldn't couldn't agree with you more strongly. We need those tools on the table. Well, thank you. I hope everyone's listening.
▶ 1:57:05Alongside the provisions in the IRA, it's imperative we continue working to invest in our nation's critical supply chains, supporting our capacity to develop and produce the high-tech revolutions essential for prosperity in this modern economy. Which is why I was proud to join my colleague Rep. Dingle in leading the Democratic Supply Chains Act last Congress. Vital provisions from this package were included in the Promoting Resilient Supply Chains Act, which was passed by this committee yesterday.
▶ 1:57:34Efforts like these not unpredictable unlawful funding cuts and across the board tariffs on our allies will help the US lead the way on AI while ensuring innovation continues to thrive in communities like mine. Mr. Turk, during your time as deputy secretary, how's the rapid growth of AI transform future planning and considerations around grid reliability and resilience? So AI is an incredibly powerful technology. uh it can help on the grids.
▶ 1:58:03Uh the grids are becoming complicated. Um uh we've got a complicated patchwork in our country. We need to not only uh have the the local grids and the regions work, but we need um uh communication, interreional uh flows if we're going to be effective uh in terms of dealing with the challenges that we've got in front of us. So AI can be an incredibly powerful tool there.
▶ 1:58:28Uh we also need an independent FIK uh to make sure that we've got good regulation, predictable regulation, regulation that has the certainty that folks can plan for uh going forward. So we need to have that regulatory environment in place too. Thank you for your response. My district is poised to lead the charge by building an innovative quantum computing campus right in Chicago southeast side.
▶ 1:58:51I'm encouraged by the promise of worldclass collaborations, exciting new technological advancements, and ensuring economic development returns to this community. With major projects like the Illinois Quantum and Micro Electronics Park and other large-scale data centers coming to the area, we must also work to bring new clean energy generation online to help meet the projected load growth in the coming years.
▶ 1:59:15We could not simultaneously pull back from these critical investments while trying to lead on AI and critical manufacturing here in the US. Last question, Mr. Turk, what critical supply chain investments need to be made to ensure that we meet projected demand while ensuring reliability and affordability? So again, we've got a whole paniply of tax incentives, grants, and loans. Let me give two specific examples. Uh we talked about critical minerals earlier in this hearing.
▶ 1:59:45Uh because of those tax incentives, because of the grant money that we've been given. Thank you for u giving us that from the Congress, from the Department of Energy perspective. Uh we've now uh made a real dent. We're on a pathway to diversifying supply chains. China holds 80% of the processing for critical minerals in our world right now. uh because of the grants, because of the loans, because of the uh tax incentives, we're on a pathway to increase in the US alone.
▶ 2:00:12Of course, working with allies, 2,100% lithium increase. I could give you a statistic for cobalt and other kinds of things. So, we're on a pathway, but this is not the time to lean back to take these tools off the tool belt. We need to lean in on that front. I'm gonna have to cut you off out of respect for my time. So, I yield back, Mr. Chair. Gentle lady yields back. The chairman recognize Mr. Weber from Texas for five minutes. Thank you, Mr. Chairman. Mr. Smith, I'm going to come to you and then Mr. Wang you next.
▶ 2:00:40When discussing the power needs of the AI industry, it's important to look at the recent track record of investment decisions in generating facilities. Constellation Energy is investing 1.6 billion with the B dollars to restart ThreeMile Island nuclear plant. Amazon Web Services paid 650 million to house a data center facility. next to a nuclear plant.
▶ 2:01:03The Homer City Generating Station in Pennsylvania is investing 10 billion with the B dollars to convert a natural gas plant. Billions of dollars of investment have gone into AI and barely any is going to wind, solar, or battery storage. So, I've got kind of a two-prong question here. Do you First of all, we'll start this way. Can you discuss Well, let me I let me make this statement.
▶ 2:01:29Is it possible that those investment companies used AI in their decision on how to invest in energy? Let that sink in for just a little bit. Okay. Can you discuss why AI? Mr. Men, I'm coming to you. Companies are investing billions of dollars into dispatchable and reliable generating resources. I know you had the conversation with Dr. done. But we're talking about real companies, real businesses making real investment decisions based on risk.
▶ 2:01:59Your thoughts? Um, so all of the data I've seen indicates that it's a fair choice now between renewables and batteries and um essentially um natural gas and so forth. In other words, the answer is you want both. How they make those decisions are highly local involving funding, permitting, processes, and so forth. Texas and and what you're doing is phenomenal.
▶ 2:02:24If you look at many of the new data centers are being built in Texas because of the environment that you all have created and some of the largest ones are being created there. Bigger than the country today. I'm I'm sorry I'm getting invaded over here by by a friendly fire. Keep going. Um, so, so, uh, Texas is a really good model of, as you know, you have your own electric grid and it's highly unregulated.
▶ 2:02:54What I like about the Texas grid is that you see real power, real pricing power on a essentially a microcond level. With respect to how people make these decisions, I hope that every company in America uses AI tools to make important strategic decisions. They are natural allies in the business decision process. Okay, I'm going to actually jump off the question line ahead. So AI, so what happens if China, who is so far ahead of us because we don't have our permitting process lined up, right?
▶ 2:03:23We're so stupid that it takes so long to permit stuff that China doesn't have that problem. What happens if they and take over our AI? What happens then? Well, I'm not suggesting they'll take they'll take over our AI. Um, our analysis is that China has very large power supplies compared to the United States. They do not have the power problem that we see. So, they can't hold our AI hostage. Um, as a technical matter, no.
▶ 2:03:52What they can do is they can do there are what are called adversarial attacks where they can essentially go in and screw with the model, excuse the term, and basically screw it up. What if they have better AI than we do? Um, that's a competitive issue. And the issue, one way to think about it is, and I I'll make an argument.
▶ 2:04:09If you and I are competitors, you're the good guy, I'm the bad guy, and I'm ahead of you, and I'm six months ahead of you, you say, "Oh, it's only six months." But if the slope of innovation is nearvertical, it's almost impossible for you to catch me up. That is a dynamically unstable. And that doesn't work when you're talking about America's security at risk. It puts our poor national security. I got you. Let me let me move on. Mr.
▶ 2:04:33weighing the energy subcommittee held a hearing on the role of AI and power America's energy future October 19th 2023 uh AI during that hearing we discussed how AI can be used to improve the performance of the grid used in oil and gas production and also some of the vulnerabilities of AI like kind of you're alluding to Mr. Smith, I have no doubt that there have been major advances in AI since adhering. So, I've got a couple questions from you. What benefits would there be from integrating AI into our nation's energy sector?
▶ 2:05:03And would you want that sooner rather than later and all the permitting to be reasonably quick? I think what what you've been alluding to through all your questions is a very uh important point which we I think have been grappling with in the AI industry which is that AI has the ability to transform nearly every industry. Um what we refer to this in the industry is how do you move towards more agentic systems? How do you move towards systems where AI are able to make more decisions more quickly and result in an overall dramatically more efficient, more effective um system?
▶ 2:05:33This will uh this will tackle every industry over time. Um but particularly in the energy sector, it's critical and the last answer is sooner rather than later. Gotcha. Mr. Chairman, I yield back. Gentleman yields back. Chair recognizes Dr. Ruiz for five minutes. Thank you. Uh Mr. Chairman, consumer protection, data privacy, and artificial intelligence impact every American. But for me, I feel a deep responsibility to ensure that our nation gets this right. Not just as a lawmaker, but as a father of two young daughters.
▶ 2:06:03I see how kids today are shaped by AI powered platforms and digital relationships like never before. While tech can inspire creativity, it also poses real risks. Studies link heavy social media use, especially for young girls, to anxiety, depression, and low self-esteem. Too often, AI algorithms promote harmful content over healthy self-worth content. Uh, Dr.
▶ 2:06:29Schmidt, you've often cited the example of an AI enabled teddy bear that learns and evolves with a child, highlighting the potential risks of such intimate AI relationships as this scenario becomes increasingly plausible. What steps are companies taking to design systems that protect rather than exploit young users? So, thank you.
▶ 2:06:59Um, every company is very concerned about the point you're making and every company is trying to address this question of, let's call it a rogue AI that comes out of themselves partly for moral reasons, but also it's just bad for business. Um, as to whether the government will ultimately regulate that area, it's not clear to me. You do have some things that you could do right now. Uh, there's a law called COPA which has a 13-year, you have to be 13 to be online. I have strongly recommended it be raised to 16 for that reason.
▶ 2:07:30Um, you can also look at section 230 and try to reduce some of the most egregious harms and that has been discussed for some time in Congress. Those small changes would take the most extreme examples of harm and take them out of the market which is probably a good thing. Thank you. You know, too often systems are designed to keep kids online longer, to collect their data and serve them ads for profit.
▶ 2:08:00They're not designed to keep them safe, respect their privacy, or ensure age appropriate content. That's why we need action. We need enforceable privacy safeguards and clear rules on how AI interacts with minors because no algorithm should decide what's best for our kids without oversight. But we must also be honest about what could stand in the way.
▶ 2:08:22The sweeping tariffs proposed last week by the Trump administration risk slowing innovation, raising costs on the very tools needed to build safer online spaces, and delaying efforts to hold tech companies accountable. They risk putting petty politics ahead of public good. and in doing so they leave our most vulnerable, our kids exposed. Dr.
▶ 2:08:42Schmidt, as efforts to strengthen data privacy and AI safeguards move forward, how do you anticipate the 2025 Trump administration's tariffs will affect our ability to develop and deploy privacy first technologies designed to better protect young users online? I don't know that I can make the com the combination in in the question. I'll have to think about it. I will tell you that tariffs are tax increases.
▶ 2:09:07Tax increases slow down progress, increase costs, lead to inflation are generally bad. Uh, Deputy Secretary Turk, the Department of Energy, particularly through its national labs, has been deeply engaged in advancing AI safety and red teaming efforts. Can you speak to the importance of DOE's role in this space and what the imple implications might be if that role were diminished or reassigned by the administration? A absolutely.
▶ 2:09:36We've got worldclass experts at our national labs, nuclear experts, bio weapons experts. Uh we need to make sure that that expertise is tapped into those individuals are utilized uh for this red teaming. Right? So before a model comes out, have those folks with their expertise working with the companies to make sure that those models not purposefully I don't think any company certainly here would purposefully put out a model that allows a terrorist to build a nuclear weapon, but they don't have the nuclear expertise to ensure that that's the case.
▶ 2:10:05That's why having these experts, these government experts, these independent experts are so important as part of that red teaming. Uh getting rid of those folks uh is absolutely a national security uh concern uh would have serious national security implications not just for AI but for everything that we need those experts for. Thank you. You know, we have the opportunity and the responsibility to get this right.
▶ 2:10:30Uh we cannot afford to wait until we see another crisis in youth mental health, another generation struggling with digital addiction or another data breach exposing millions of children's personal information. So I urge my colleagues, let's put families first. Thank you and I yield back. Thank you that Dr. R yields back. Mr. Allen's recognized for five minutes.
▶ 2:10:50Thank you, Chair Guthrie, uh for hosting this full committee hearing examining AI and impacts uh had in impacts development with uh an allout energy approach and the technology and uh certainly we've got to be competitive globally.
▶ 2:11:08Uh in fact uh you know we wouldn't be hiring this we wouldn't be having this hearing if President Trump were not in office because he has said that his part of his agenda is an allout energy uh program.
▶ 2:11:22uh we need to uh dominate energy in the world and of course AI uh we understand that that race and uh uh so uh and everything that uh we're doing our conference is doing is to uh provide every opportunity for us to be the energy dominant country that we were just six years ago.
▶ 2:11:45And so that's what's so critical and that's why we're having uh this uh hearing today to find out okay what do we need to do to make that happen. I'd like to thank you for being here. Uh our witnesses uh you know with the emergence of AI the US has to be a global leader. Uh to be a leader in AI it is critical that our energy sector is equipped to meet the demands of that AI poses. Dr.
▶ 2:12:13Smith in by 2030 data centers can consume upwards of 9% of total US electricity at the same time as we're seeing historic projections of electricity demand because of AI developments in the nation's bulk power system is already under incredible strains. In fact, the North American Electrical Electric Reli Reliable Corporation or NERK found in their last long-term reliability assessment that half the nation is at risk of resource adequacy. We know in some states we're having brownouts.
▶ 2:12:43That is half the nation is at risk of blackouts during times of extreme weather. In my opinion, our nation will need significantly more power to meet these demands and fast. How can we balance the needs of everyday Americans to keep the lights on while simultaneously powering developments in AI models? Um, the answer of course starts with our overall premise which is more of everything. It also includes a more intelligent grid that is more flexible when bad things happen. that is now possible with AI and with grid modernization.
▶ 2:13:13You need both. Uh Mr. Patilla, last month we heard directly from the grid operators talking about grid who are charged with overseeing the reliability of our electric grid and they highlighted one of the biggest issues facing the bulk power system is the premature retirement of base load power plants which has been mentioned quite often in this hearing.
▶ 2:13:32We also heard that places like New England who don't who do not have sufficient natural gas capability and long-standing opposition to nuclear energy are not seeing the same uptick in new investments or data centers and manufacturing facilities. Uh my home state of Georgia which has been the best state to do business in 12 years in a row has been a leader in investment in job creating industries largely because of our pro business environment and diverse uh slate of base load generating resources.
▶ 2:14:01In fact, now we we have just added to Plant Vogle uh two more units and it is the largest clean energy facility in the United States built in the last 30 years. Given your company's energyintensive nature, how important is access to reliable, affordable electricity when deciding where to invest in US manufacturing facilities? Thank you, Congressman.
▶ 2:14:29And I'd like to just start by uh giving a a call out to our research and development center that we have in your home home state and uh uh we have we have certainly found that over time that that has been a a wonderful place for us to attract talent and grow our our engineering capabilities there. In terms of your question on base load absolutely I think many of the questions today have been focused on that and for you know semiconductor operations we have very very consistent loads.
▶ 2:14:56We have of course high loads and the the reliability of the power is incredibly important as I've mentioned earlier. So nuclear power, hydroelectric power, these are excellent um fits for us. But we also agree with the other panelists and all of the above approach is what's is what's required. Good. Uh I have a few uh Mr. T I have a few yes or no questions I'm going to ask you. Uh do you agree that permitting reform is needed to meet as you discuss in your testimony our rising energy demand? Yes.
▶ 2:15:26Uh, do you agree that it includes air permitting? Yes. Uh, the national ambient air quality standards implemented by the Bar Biden Harris administration's EPA, for example, have proven to be a significant burden on the US manufacturing base. These stringent regulations have made it difficult to permit and develop many of the facilities needed to support our next generation of industrial base. Whether it be PM2.5 or ozone, EPA needs to be more flexible.
▶ 2:15:55No question about it. Your time's uh my time has expired. I have an additional question for you. If you would answer that for the record, I would appreciate it. And I yield back. Thank you. Gentleman yields back. Chair recognizes Miss Clark for five minutes for questions. Thank you very much, Chairman Guthrie, Ranking Member Palone, and to our panelists uh for today's hearing. Thank you to our witnesses for being here to testify.
▶ 2:16:20And let me just say that it is a pleasure to see Micron presented uh represented on this panel as Micron is making historic investments in New York that will transform our state and the semiconductor industry more broadly. Members of this committee are well aware that generative artificial intelligence has proven to be one of the most impressive technological advancements of this generation.
▶ 2:16:45But with a tool so expansive, it is up to us to ensure that AI systems are developed and deployed responsibly and with consumers in mind. Last Congress, I had the honor of being appointed to the bipartisan task force on artificial intelligence, which was established to ensure that the United States continues to lead in AI innovation, as well as examine appropriate guard rails to protect against emerging threats like those outlined in the 2023 GAO report on the rapid
▶ 2:17:15use and growth of AI. I've been sounding the alarm on issues related to AI and algorithms for years, namely the potential for algorithmic bias. AI has only gotten smarter, and with its rapid development, consumers are faced with the increasingly acute potential for harm caused by algorithmic discrimination.
▶ 2:17:36For example, facial recognition technology, a tool used by both retail stores and law enforcement, has repeatedly shown an inability to accurately identify people of color, which has led to multiple instances of false identification and unwarranted harassment. And when it comes to home ownership, black applicants are denied mortgages at higher rates, a decision that is increasingly made based on algorithms.
▶ 2:18:03In healthcare, al algorithmic bias can lead to misdiagnosis as a people of color are historically under reppresented in existing data sets and algorithms are improperly tested for accuracy. My top priority with respect to the growing use of AI is simple. We need to make it abundantly clear to developers and deployers of algorithmic systems that Americans do not forfeit their civil liberties when they go online.
▶ 2:18:31That is why I have prioritized algorithmic accountability and have fought to codify and make explicitly clear that civil rights protections still apply in the digital realm, especially when AI is used in critical decisionmaking. Lines of code remain exempt from our anti-discrimination laws and too often go unchecked. Every algorithm has an author. Every bias has an origin.
▶ 2:18:58Through proper regulation, we must ensure safety, inclusion, and equity are top of mind in the deployment of automated critical decision-making systems that affect Americans lives. And while I'm pleased with the final report of the bipartisan task force on AI and find that it serves as a productive framework to set guard rails on AI that includes civil rights and liberties, the conversation does not end there.
▶ 2:19:25It is up to this committee, my Republican colleagues who are seemingly have an aversion to the word civil rights to properly protect all Americans when they either electively or unknowingly use AI to make critical life decisions. I have one question, Mr. Turk. Do you agree that it is important to ensure that AI systems are rigorously tested for bias before they are deployed and on a regular basis thereafter?
▶ 2:19:55Well, let me first thank you for your leadership on the bipartisan task force and more generally and uh I completely agree. We need to have those kinds of protections in place. This is a powerful technology, incredibly powerful technology and we need to get this right.
▶ 2:20:08Yeah, I'm just concerned that, you know, some biases get get baked into our systems and that inaccuracy can be uh detrimental not only to communities but to our ability as a nation to be as strong as possible, particularly when guarding against um adversaries that seek to do us harm. So, thank you for your work, gentlemen. I appreciate all that you're doing. Young man, Mr. Wayne, you're making it happen. We're proud of you.
▶ 2:20:37much continued success and I yield back the balance of my time. Thank you. The gentle lady yields back and the chair recognized Mr. Balderson for five minutes. Thank you, Mr. Chairman, and thank you for holding this uh hearing today. And I'd like to thank all the witnesses for being here also today. Uh let me start with you, Dr. Schmidt. Uh I represent Ohio's 12th congressional district, which covers central and southeastern Ohio. Licking County, which I'm proud to represent, has one of the largest clusters of data centers in the country.
▶ 2:21:07Google, Amazon, Web Services, Meta, QTS, Vantage, and more all over data centers in central Ohio and my district. In total, power demand from data centers will reach 5,000 megawws in central Ohio by 2030 based on sing signed power agreements. Just last month, Williams announced a $ 1.6 6 billion investment to build new two new natural gas fire plants in Licking County with a combined capacity of 400 megawws.
▶ 2:21:35This reliable base load power generation is critical to meet growing demand in central Ohio. Dr. Smith in order to alleviate strain on the electric grid. I'm curious what role or involvement you think these tech companies should have in helping to bring in new generation to secure the massive amounts of power needed for their facilities and how should these companies partner with grid operators or power providers to ensure we can properly account for tracking growing tracking demand.
▶ 2:22:03So, uh, when I was at Google, we made a bet on Ohio and we built the largest data center at the time in the world, which was massive. And I used to go visit it and say, "Oh my god, the data centers you're describing are 10 times larger than anything I ever built way back when I was doing this only seven years ago." So, it gives you a sense of the scale of the investment in what you're doing. The best thing to do is to have a strategy within your state where everybody agrees to solve the energy power problem.
▶ 2:22:34We found in working in Ohio that we were able to get access to the high voltage lines that we could not get access elsewhere. We built our own substations which are also massive. That's what it takes. That's what every one of you is going to have to do to have your states be a center for AI, the AI revolution. Thank you. Um Mr. Batia, I switch to you. I'd also like to hear your thoughts on this. What is Micron doing to be proactive in securing the power needed for these chip fabs?
▶ 2:23:04So um as part of our uh selection of the locations where we will be expanding um the power availability and the uh uh the um agreements that we could reach with local power companies was a key part of that criteria.
▶ 2:23:18As I mentioned before, um nuclear power, hydroelectric power both very good fits for us and um uh and those are uh in strong um availability in the areas where we selected and we continue to work with the providers in those areas to be able to ensure that we can have more investments to be able to have uh long-term access to that affordable and uh reliable power. Thank you. I I'll continue with you, sir.
▶ 2:23:43Um, you note that one of America's strongest competitive advantages compared to markets in Asia is our reliable and affordable energy supply. I strongly agree with you with this assessment that we must maintain this key competitive advantage by building out generating capacity to meet the expected short-term surge in energy demand after years of flat growth. However, right now we're seeing massive backlogs of generation project and grid operators interconnection cues depending upon the region.
▶ 2:24:12power projects are sitting and waiting in interconnection cues for five years before they can even get studied and then ultimately built and connected to the grid. The buildout of AI and data centers isn't happening in five years. It isn't happening now and these facilities need power. Do you have concerns that the current process can take years and years just for new power generation projects to get through the Yes. Thank you.
▶ 2:24:39Do you think Congress can play a role in ensuring new generation is getting online and connected faster given the historic increase in power demand? Yes. Thank you very much. Dr. Schmidt, would you like to add anything to that? The interconnection cues are a very good example of something which is something that you all need to work on. Uh basically getting the system to be more flexible when the industry shows the demand. I mean the the delays are crazy, right?
▶ 2:25:06people, they have the money, they have the ability to get the power built, and they can't interconnect it. That's a good example of grid modernization. It applies to everybody. So, I encourage you um we introduced some legislation called the grid act, and uh it's all about the interconnection cues. So, with that, Mr. Chairman, I yield back. Thank you. Gentleman yields. The chair recognizes a gentleman from California, Mr. Peters, for his five minutes. Thank you, Mr. Chairman. I do think it's great that the whole committee is hearing this and I thought particularly Dr.
▶ 2:25:34Schmidt, I want to compliment you on setting the um setting the the table on this because we we're caught up in a lot of little things and you really gave us a very big perspective on how important and daunting this is. Um I had a bunch of questions from before. I'm only going to ask one which is about um the energy permitting reform act or EPRA at the which was the Senate permitting deal at the end. Mr. Mr.
▶ 2:25:56Turk, can you talk a little bit about the importance of transmission and the importance of re interreional planning and interregional transmission as a way to help deal with our energy needs? Well, put put simply, we just need to make improvements. Um, we need to get the most out of our existing transmission uh grid enhancing technologies, reconductoring, bringing AI and other tools to make sure that we're uh smart about these assets and they're flexible and they're smart.
▶ 2:26:22Uh and then secondly, we do need new transmission and it does take too long in our country to build transmission. Uh we under uh went a whole uh series of reforms in the Biden administration to try to improve that and I think we made some significant progress, but we need to do more and certainly Congress has an incredibly important role here. Well, I'm an advocate on this committee and the energy subcommittee for permit reform. We we did a lot of work to get EPRA to the point it was. I think we should start with that um and adopt it.
▶ 2:26:48Um I I would say that this concept of um all of the above energy I I understand I think sometimes it's all of the above as long as it's natural gas. Um the thing I would say is we I was recently at a meeting with the energy company and some of my Republican colleagues where they said now it takes five years to get a national natural gas plant online largely because of the supply chain constraints and getting turbines. Um but you can get solar within a year.
▶ 2:27:15uh and the the companies was begging us not to repeal the incentives for solar power uh and probably wind um that are in the IRA and I hope when we talk about all the above we are really committed to all of the above and that we don't do something to to shoot ourselves in the foot with respect to natural gas I'm more than willing to work on natural gas I've been saying like a broken record what I want out of that from an environmental perspective is some agreement on the regulation of fugitive emissions methane emissions It's an easy thing
▶ 2:27:45to do. It's something the industry is open to. If we did that here, it would would solve a lot of the it would answer a lot of the questions we have about uh the use of methane um use of natural gas as a bridge fuel. I think that's an easy thing to follow. I would reiterate what some other people see. The need to invest in basic science is really critical here. Uh China is is is is more than keeping pace with us. They're out investing us by quite a bit.
▶ 2:28:10uh in my district a lot of that is um is um in biotechnology but a lot of it is in uh fusion and I think that's something that we have to continue to invest in uh it's it it would really solve a lot of problems but it takes investment and I think investment in energy in our universities the best university system in the world best set of universities in the world is really critical this for this country I agree that our data is is ridiculously unmanaged and um uh and uncoordinated we saw this in COVID.
▶ 2:28:39I tried to deal with it in COVID. You can't draw conclusions from a data set that's so disperate and unorganized. I think um that that your comments were really wise about that, sir. And I think that um Congress has a role in making sure that we get on top of that. Um I I would do not don't want to um overlook the role of imports in this. I mean, we do not make solar panels uh here. We import a lot of things. We're making it more expensive. It's craziness. It's a craziness.
▶ 2:29:06And I think for the Republicans who used to be such staunch supporters of free trade and Democrats like me who supported both the Trans-Pacific Partnership and the USMCA negotiated by Donald Trump, uh we can't forget the benefits of that international trading system and that the cost of inputs that we need to solve this problem are really being heightened by this trade war and this self harm.
▶ 2:29:32Um I do want to say too that um I think we should think hard about whether some of the calculations can be decentralized. Um we've we've we've we've um we've taken for granted we've taken as a given that there's a certain amount of energy we need. I have no doubt that it's a lot. I had actually heard five gigawatts for a data center. Now I hear 10. Uh that's a massive challenge.
▶ 2:29:54we ought to think about whether as a systemically some of those calculations could be done on these handheld devices would take some of the power uh requirements away from those big facilities. Um and finally I would the other thing I would observe as a Californian is we can't let ourselves get into the situation we're in with privacy where we have 38 different standards across the country. This committee has got to come to to grips with the notion we have to do preeemption. There's a federal supremacy clause for that reason.
▶ 2:30:24This has got to be a national policy. We've got to set national standards. We got to do it on privacy. We got to do it on AI. And we can't be scared of using our power here. Uh thank you very much. We have a lot of work to do. Again, thanks so much for the witnesses. And I yield back. The gentleman yields. The chair recognizes the gentleman from Texas, Mr. Fluger, for five minutes. Thank you, Mr. Chairman. Uh Dr. Schmidt, Mr. Um, what role will LG play in providing the power that's necessary for AI and data centers? I'm sorry, LNG.
▶ 2:30:54Natural gas. So, natural gas. Um, it sure looks like natural gas is needed in most renewable scenarios because of essentially a peaker plant. It also looks like uh we just need more natural gas, more natural gas generation kind of everywhere. Mr. brought you you mentioned in your testimony earlier today about uh concerns with intermittent power. Uh so when we're looking at sourcing, what what do you need intermittent or do you need reliable base location?
▶ 2:31:24Well, so we have obviously um uh we need reliable uh base load power um natural gas has the ability to be able to be a really good smoothing capability for the and buffer basically for the ups and downs of the overall grid. And that's why I would agree with Dr. Schmidt that it's an important area and it's an element that it's an area that the United States has, you know, a tremendous amount of of capability in.
▶ 2:31:48The um power providers were here testified last week and they said that Texas alone is at a peak demand of about 80 to 85 gigs right now and that's going to increase in the next four to five years to 150. So Mr. Turk. Um, are you familiar with the study that DOE did last year? Uh, actually started in 2023 uh on LNG. Uh, I am. What was your role in Department of Energy? I was at that time the deputy secretary, the number two official.
▶ 2:32:19Okay. Did you and I was very involved? You were involved with that report? I was. What was the title of that report? I don't remember what the title of the report was. Uh what we did is we asked uh a number of our national labs to give us an independent assessment. Okay. Um when was that report released? Uh we pushed our national labs to do it as quickly as possible and when did department report I think we ended up releasing it late uh last year early this year. Okay.
▶ 2:32:46So you actually uh you actually did release it. We did release it. Were there sections that were redacted? Uh not to my knowledge. Uh we believe very strongly we we wanted an independent analysis to look at the cost implications, the environmental implications and uh we did not suppress any information whatsoever. Uh were you aware of the 2023 studies findings prior to the January 26th decision to indefinitely ban new export authorizations under section 3 of the natural gas act.
▶ 2:33:16So we didn't ban any uh any we did the study in order to take a step back because we've authorized so much up to half of our natural gas production right now is authorized to actually go abroad and to be sold including to China. So what we did was take a pause the study I'll reclaim my Mr. Ch. Thank you. Thank you. Um pause and we can debate this all day long. Um but why was the study not released immediately after it was done?
▶ 2:33:44So it was um we released the study once the experts uh finished the study. Do you disagree that uh the study um was more favorable to LNG than the Biden administration would have liked and that's why there was a pause put on LG exports. The study the pause was so that we could do the study before making decisions. So um and to actually have our independent experts and uh uh the independent experts in our national labs were the one who uh did the study. Okay.
▶ 2:34:15So the study actually came out was released by Secretary Wright um and uh so we released the study the Biden administration released in December of 26 um or December of 24, excuse me. Um, and it came out as a pretty favorable uh with regards to emissions. Um, but it was delayed by the Biden administration for months. It wasn't on being released. Well, that that's it wasn't I I was there. It wasn't delayed.
▶ 2:34:42It that's how long it took because we wanted a thorough independent analysis by several of our national labs. So, do you agree that the emissions uh of natural gas um were were better and more consistent and actually more favorable than what you claimed and what Secretary Granholm claimed in the attempt to ban natural gas exports? So, LG exports have a very very significant n uh uh very significant greenhouse gas footprint.
▶ 2:35:10So, just one project uh we're talking 4 BCF per day. uh that project itself has more emissions throughout the life cycle methane emissions but CO2 combustion when that gas is burned than 141 countries in our world one facility 141 countries in our world that's a pretty significant so you stand by your decision to ban LG exports again we did a pause so we could do the study so that any secretary of state could have good independent analysis your decision to do that is going to impact
▶ 2:35:40these guys right here it's going to impact our ability to provide power for for the AI data center. So again, that's LNG that's being exported. This had nothing to do with domestic use of gas here. Mr. In fact, the more we export, the more price pressure for our the gentleman yields domestically. The chair now recognized Mr. Sodto from Florida for his five minutes. Thank you, Mr. Chairman. You know, listening to this hearing, I feel like we're in a time warp back to 2024. Biden was president.
▶ 2:36:08We had the strongest economy in the world and we were free to debate the finer points of AI regulation, the IRA and data centers. But it's April 9th, 80 days into the Trump administration, and Trump's tariffs, chaos, and deportations have sent our economy into a freef fall. Will our friends across the aisle just bury your heads in the sand and pretend this isn't happening? Or will you join us to help fix it?
▶ 2:36:32Speaker Johnson just today blocked any consideration of tariffs until September 30th. He put a straight jacket on the US House of Representatives to even try to address this issue. Meanwhile, AI data centers could see an estimated 30% increase in expenses to build.
▶ 2:36:53According to Fortune magazine this week, air conditioning, liquid cooling systems, transformers, circuit breakers, cabling, routers, switches, construction materials, battery systems will all go up because of Trump's tariffs. So, is the biggest threat to AI overregulation? Or is it the tariffs? Duh. Mr. Turk, what do you think is the biggest threat right now to AI development? Is it the overregulation allegedly or is it tariffs?
▶ 2:37:24I think tariffs uh increase costs and they increase uncertainty and that is damaging for AI being built in our country but it's damaging across our economy. And then we I'm worried about demand and access to capital. are heard in both Newsweek and Fortune mag magazine this week. Access to capital is in real jeopardy because major tech companies, the biggest investors in AI see a potential recession on the way and their core businesses are threatened.
▶ 2:37:51Ad spending drops, the capital drops during a recession. Uh Dr. Schmidt, we saw that the Google shares were at $200 a share when Trump took office and now they're at $146 a 27% drop in three months. No one celebrates that. That's awful. That would have a negative effect on future AI investments for Google right now. Isn't that true? Don't remind me the stock price. Yes, I didn't mean that. I'm not here to attack anybody.
▶ 2:38:21But that but how does that affect Google's investment in future AI? I can't speak for Google, but I can say in general the genius of the American financial system, aside from the fact that we're a reserve currency, is that crazy entrepreneurs can raise billions of dollars on a whim, on a risk. That's why we're leading. If that system breaks, the break the system that is the unification of the government, the private sector and academics and that money is not available, we're toast.
▶ 2:38:51Yeah, we're the world's currency right now, but you know that's in jeopardy as we speak. Uh Mr. Batia, we saw Micron take a thump too from 109 per share when Trump took office to 65 today, 41% drop. Again, no one likes this or celebrates this, but how does that affect your access to capital and the ability for you to continue to develop AI chips, AI microchips in technology?
▶ 2:39:17So, you know, we have uh um we take a long-term view and the demand for growing for memory, the demand for data. Therefore, the demand for memory continues to grow. It's a secular trend. And so, we we we um intend our investments to be for the long term, but we have to bring them online in line with the the the demand trends that we see. And so we we continue to expand in that way. But that um that shows the importance of uh and these kind of uh uh volatile events will happen from time to time in our industry.
▶ 2:39:44And that shows the importance of us having um a durable predictable investment tax credit to be able to support um our continued expansion here in the United States where we're where we're committed to building. We all want to make sure these stocks go back up and people's retirements are protected. And that's why this Congress needs to work together. We did work together on the bipartisan advance act which boosted uh nuclear signed by President Biden bipartisan product from this committee. U Mr.
▶ 2:40:11Turk how does the speeding up of deployment licensing and new reactors and fuels help u through nuclear the future of AI. So I think it's a big deal and thank you for the leadership on the advance act. Um we need to not only get the most out of the resources that we've got including those resources that can be brought on quickly to our grid.
▶ 2:40:30right now that's solar and storage uh and wind uh those are the resources that allow us to bring electrons on quickly to power these AI data centers but we absolutely have to work on clean base load power nuclear is an incredibly important part of that equation enhanced geothermal is another incredibly important part and so we need to have the research we need to have the investment and we need to have those tools as quickly as we can thanks I yield back gentleman yields the chair now recognizes a gentleman from Idaho Mr. Fulture for five minutes. Thank you, Mr. Chairman.
▶ 2:41:01Mr. Petilla, in a different era back years ago. I was privileged to work for Micron Technology. Technically, I began with the startup phase. It was still in startup phase. Spent 15 years there. I cannot tell you how proud I am of that. And the education that I received personally, the experience, life experience, that uh was truly unique and I could not have higher regard for your company.
▶ 2:41:25And so please know that at that time as I say things were different but we sold ourselves we we positioned ourselves with customers and potential customers that that uh that most of our costs were fixed and whether we produced one die or a million die the costs were largely the same and that gave us the ability to sell ourselves as American supplier.
▶ 2:41:53Now today you discussed how energy is one of the factors that's changed that business model and back in those days it was all about die size and could we stack the capacitors and make it efficient and that was the secret sauce and if we got that we won. What's changed in today's business model other than the energy that you correctly spoke about to change that uh strategy and business model in the framework that you're operating in today?
▶ 2:42:23Thank you. Thank you, Congressman, and thank you for your time in the early days of Micron and helping put the company on a on the on the track to where it is today. Um I think the biggest thing that's changed is the cost competitiveness of building and operating fabs uh in the United States over over this last 25 to 30 years has become a widening gap between um doing that in the United States versus uh Asian countries where we operate and where our competitors construction's cost just to be clear construction construction construction cost is one of the biggest
▶ 2:42:53gaps it's probably the biggest gap between the Asian um uh countries and where our competitors are versus uh versus United States. Um the uh in fact the energy is an area that has been a bright spot for the United States and it's an area that you know focus of this hearing is to make sure that it continues to be an area of advantage um for uh for semiconductor industry for Micron but also for many other industries so that we're able to be able to make sure that all of these projects can come to fruition
▶ 2:43:25and workforce and the the the investments we're making in workforce. You know, we certainly believe that in partnership with um the uh many different universities that we're working with across the country that we're going to be able to redevelop a pipeline of skills and capabilities that have been lost over the years as manufacturing left the country.
▶ 2:43:44We're also working with various different um military exit organizations to be able to to train veterans to come and work because we think there's a really good uh overlap between the skills that they have from the military and the skills that they have to be able to operate and maintain uh fabs. And I think um you know as I've mentioned before um expanding and extending the ex the currently expiring investment tax credit for semiconductor uh projects is really really very critical. Got it. I'm going to come back to you if I have time. Mr. Wang, I want to ask you a question.
▶ 2:44:15In your testimony, you said there's three things Congress should do in order to move forward regarding AI. The second point you made was one, federal AI standard. We've had discussions about that on the committee. Dig that a little deeper. Peel that onion back. Specifically, what should those standards be to the best of your advice? Yes. Um, so first of all, uh, just speaking as an AI company and being able to operate and innovate effectively, we need we need one federal standard.
▶ 2:44:44We cannot afford a patchwork of 50 different state standards that we have to execute against. Um, then peeling the onion back, we believe that we need to create this uh, a regulatory framework that enables innovation while still uh, adding some level of guardrail.
▶ 2:44:59So our view is we need a use case sector specific regulatory framework where in certain industries like medicine or financial services or insurance or others where there should be heightened uh levels of scrutiny or heightened levels of controls for what AI systems uh can and should do um we should put those in place but in other industries where we want uh the core technology to advance more rapidly and more effectively we need to allow that to happen.
▶ 2:45:25Okay, I'm about out of time and I'm going to submit some questions for the record, uh, Mr. Chairman, but Mr. Wang, I I I do think you're a wealth of knowledge and I would just say to you and the rest of the committee as we go about forward in setting or trying to set some federal standards, please be careful what you ask for because you just might get it. And we can be a friend or we can be a very ugly big brother. And I I say that because it's very difficult to identify the proper role of federal government with these things.
▶ 2:45:55So thank you to all those who testify. Mr. Chairman, I yield back. Gentleman yields. The chair recognizes the gentle lady from Michigan, Mrs. Dingle, for her five minutes. Thank you, Mr. Chairman, and thank to Chair Guthrie and ranking member Palone for convening this hearing. AI, as has been discussed today, is transforming every sector from healthc care and transportation to manufacturing.
▶ 2:46:18But with rapid advancements come serious challenges such as data privacy risks, algorithmic bias, and the growing threat of foreign adversaries exploiting our vulnerabilities. And we cannot afford to let America's data and personal information be weaponized by China or other adversaries or allow allow AI to spread unchecked through deep fakes, robocalls, and deceptive ads.
▶ 2:46:44So, I was proud to help lead Take It Down Act, which passed out of the committee yesterday to hold bad actors accountable for sharing non-consensual deep fake content online and to protect survivors. AI when paired with 5G and emerging technologies, is already transforming lives, streamlining public services, modernizing transportation, and improving health care outcomes. But to lead, we have to invest.
▶ 2:47:13That's what we were doing during the Biden administration. And quite frankly, I'm very worried that we are now witnessing efforts to undo that progress. Programs that were signed into law through the bipartisan infrastructure law, the chips and science act, the inflation reduction act, all of which provided funding that is crucial for the AI e ecosystem are being dismantled. These investments aren't just about clean energy. That's what people don't understand.
▶ 2:47:41and they're about global competitiveness, job creation, and securing the future of the American industry. The IRA has been critical to accelerating domestic man manufacturing, especially in the auto sector. I admit that's one I care about deeply, which remains the backbone of the American economy. My Republican colleagues say we must out compete China in AI. They're right. We must. I agree. But you don't win it by slashing your own tires.
▶ 2:48:09You can't lead by cutting funding, firing experts, and abandoning the public private partnerships that fuel innovation. Secretary Turk, does cutting funding from agencies like the Commerce Department's Bureau of Industry and Security, holding up chips, investments, threatening that they may not happen, firing technical experts, and NIST keep the US competitive in the global AI race, especially as China ramps up its investments. And what happens if we walk away from chips and the IRA incentives?
▶ 2:48:40I think this is exactly the wrong time to walk away for those from those Okay. Short. Mr. Batia, what are the I like it. It's wrong. What are the consequences of repealing the tax credits and public investments that are driving domestic industry growth in clean energy and advanced man Well, Congresswoman, first I'd like to just u uh comment that I'm proud of and born and raised in Michigan.
▶ 2:49:08My my first job in manufacturing was more than 30 years ago in the body shop and that created my my uh love of manufacturing and you know a thriving automotive industry as you said in your comments is I think critical for the uh for the country's um you know economic health as well as for national security. um absolutely agree that we need to have continued uh support for investment tax credits for areas that are critical to AI, including of course semiconductor manufacturing.
▶ 2:49:35Um uh the the uh tax credit that was passed um is expiring and this will uh create a challenge for continued investment uh especially long-term investment because this is not just a 5-year race. This is a 15 20 year race that we're we're uh getting into and we want to make sure we have leadership in technology and capacity together to be able to um lead in uh in creating the in enabling the the AI uh revolution. Thank you. Dr.
▶ 2:50:03Schmidt, do companies operating in the US currently have meaningful incentives to protect consumer data and privacy? Are the current patchwork of state laws and voluntary standards sufficient? or would a comprehensive federal privacy law with strong data minimization provide greater clarity and consistency for both consumers and the industry? I think there's a general view in the industry that a single privacy law would be a good outcome. I think it'll be very hard to achieve.
▶ 2:50:34My own opinion is given that's hard to achieve, you're better off working on the most extreme cases such as I fully support the bill you did yesterday. That's a good example of an extreme case. Maybe there's some other extreme cases that we could also handle through your good work. Well, more questions and I want to dig into that too, Mr. Chairman, because I'm out. I'm going to have questions for the record as some of my other colleagues do, but this is a very important issue. All of them are. Thank you, Mr. Chair, and I yield back. The general lady yields.
▶ 2:51:03The chair recognizes a gentleoman from Tennessee, Dr. Harshburgger, for her five minutes. Thank you, Mr. Chairman. Thank you to the witnesses for being here today. I'll start with you, Dr. Smith. Uh, when tech companies are building the future of AI uh, in the United States, we know these data centers use massive sums of energy and for the most part they're going to be running at maximum uh, capacity 24/7 and this technology requires more base load power production rather than
▶ 2:51:33renewables like wind and energy uh, where that production fluctuates. And my question is, how would it strengthen um America's bid to lead the AI economy if we adopted a more friendly environment for natural gas and build out additional uh pipeline infrastructure? Um I agree with the need for more natural gas in the United States, more pipelines. I'd also point out that you can achieve the same base load goal with a combination of batteries and renewable.
▶ 2:52:02I think that the industry and the energy suppliers should make those on an economic basis. And I think the collective panel here is telling telling all of you all of more is better. Yeah. All of the above. Yeah. Um Mr. um Via, your testimony goes into great detail about the difficulties of navigating US permitting law.
▶ 2:52:25Uh does the challenge Micron faces when building a facility like the one you're working on in New York um chipmakers, would it make the chip makers reconsider the United States? and if so uh how could the US be could it be leaving opportunities on the table by failing to update NEPA?
▶ 2:52:45So certainly um we have uh you know experienced um delays and and um the uh the duplicative nature of of the process has um been a challenge and it's a challenge for uh I think any company who has to go through the NEPA process whether in semiconductors or in other areas and there will be other NEPA projects including in potentially in energy and in in other sectors where um you know I think that there is a potential for uh some streamlining to uh have
▶ 2:53:15federal and state processes to be um harmonized so that we don't have to go through the extended time. Yeah, it is duplicative. I mean very much so. Thank you, sir. Um Mr. long. One thing I really loved is government efficiency and uh I was inspired by your testimony uh by your recommendation that implementing AI uh applications for the government could free up public employees to think more strategically and could reduce regulatory backlogs.
▶ 2:53:46So, how could the administration use AI to lower taxpayer burdens and increase government efficiency? the opportunities for AI to aid in government efficiency are are immense and this is one of the areas where I think AI can have tremendous impact uh very very quickly actually um you know this goes to one of the things that that we're talking a lot about in the industry which is moving towards an agentic government so how can we enable AI agents to start uh speeding up and and and
▶ 2:54:16streamlining a lot of the processes that we have within the government so that they go from from years to weeks or potentially even days. So, a few examples of that, you know, I think about um uh how we can use AI to cut down the time it takes to handle veteran healthcare paperwork um or an AI system that could vastly improve fraud detection at the IRS. Um and then, you know, I think I think the combination, you know, if you look at every single agency, there's immense opportunity.
▶ 2:54:43Um and you can go across, you know, we see this in the DoD who we work very closely with. Um we're working with them. Uh recently we've been deploying a system called Thunderforge which is a uh uh system to using for using AI for military planning and wargaming a process that currently is extremely manually intensive and we all know that to be competitive in the future we need to be more efficient.
▶ 2:55:06Um, so there's just an a wealth of opportunity, which is one of the reasons why we recommend that ideally every federal agency should have some flagship AI programs to start implementing and getting into the process of utilizing AI and AI agents to uh to streamline more of their processes. And ultimately, if we do that today, we'll we'll we'll reap the benefits in the years to come. You're right.
▶ 2:55:28I see that already in some of the things we've already found um with fraud, waste, and abuse and some other we don't even communicate within an agency for heaven's sakes. So, AI would absolutely benefit. You keep doing your work, young man. Okay. All right. With that, Mr. Chairman, I yield back. General lady yields. The chair recognizes Mr. VC from Texas for his five minutes of questioning. Uh thank you, sir. Uh and I think this is amazing that we're here having this conversation today.
▶ 2:55:57Uh right now uh in the Dallas Fort Worth region, we are literally becoming a hub for advanced manufacturing and AI innovation. Uh and you can see it in all the new facilities that are opening up. Um you know, we've had the Facebook data center for a long time now, but we just we we have groundbreaking on several other new centers in the Alliance corridor near Crowley near and near Benbrrook and Fort Worth.
▶ 2:56:20Uh and this is really amazing because you can see the new facilities going up uh and the jobs that they're bringing along with them which is very very important. Uh it's generational uh and it's really helping the DFW area lead the charge in this area. But as we lean into the future we have to be cleareyed about what comes with it. Uh because AI just doesn't run on code and it it runs on power.
▶ 2:56:42uh something that we talk a lot about uh on this committee and with a massive growth of data centers and AI infrastructure uh that is happening right now is putting a tremendous strain uh on the grid and we have to get ahead of this or or consumers could end up uh footing the bill on this through higher prices and tighter capacity and more volatility and so we can't treat energy demand from AI like an afterthought. Uh we got to be smart. Uh we got to keep the lights on. We got to keep the bills affordable.
▶ 2:57:11Uh and we got to keep the grid resilient. Uh especially in places like Texas. Uh that's a huge part of our country's uh economy. Uh but also because we've already seen by by what not investing in the grid can look like in 2021 during wintertorm Yuri. uh because uh if AI moves forward without guard rails uh for jobs, for privacy uh and for families, uh we really risk turning a lot of this promise into disruption.
▶ 2:57:41And I had a question for Mr. Turk. Uh do you believe the Department of Energy or Congress needs to take more aggressive steps to plan for and manage the energy load coming from AI infrastructure? And are there policies that you'd recommend uh to ensure grid reliability uh and again Uh absolutely is the short answer and uh fortunately Congress provided a whole range of tax incentives, grants and loans uh that are having
▶ 2:58:11a real impact right now on making prices more affordable not only for AI companies but also for consumers across the country. And it's helping to improve our grid reliability. Also, I know there's an active discussion going on right now in Congress. Uh, do you all repeal those tax incentives that are lowering costs and allowing us to bring electrons on more quickly? And we look at uh what type of electrons are going to be brought on more most quickly in our country. Uh, it's solar, it's wind, it's storage.
▶ 2:58:41That's what the experts, that's what the utility CEOs are saying. Uh, unfortunately, right now, we have a uh backlog on natural gas turbines right now. that is making it very challenging to bring natural gas on as quickly as some AI companies might want it to. So if you want to bring on electrons quickly, uh keep those tax incentives, keep those grants, keep those loans in place so that we can do it quickly, we can do it affordably, and that reduces costs for everybody, including for consumers. Yeah, absolutely.
▶ 2:59:10And it keeps America ahead by us investing in those things. Uh you were at DOE when the chips and science act passed um a law that's helping bring semic bring semiconductor and AI related manufacturing uh back to US soil. If those incentives are rolled back, do you think companies would continue to invest in domest domestic manufacturing or would they uh move those operations overseas? I think Dr.
▶ 2:59:35Schmidt described what happened unfortunately uh a decade two decades ago when we let those manufacturing facilities slip out of our hands and go to other countries and the chips and science act was Congress working with the administration to step up and say we need to bring that back and it's going to take some upfront capital. It's a perfectly appropriate role for the government to say this is a critical technology. We're going to invest we're going to encourage and sent a bunch of private sector investment to have those chips uh manufactured here in the US.
▶ 3:00:05So the short answer to your question is if we were to some reason slow down the CHIPS act or resend that funding uh we're going to be right back where we were which is not where we need to be. Damn. If for some reason we didn't fully implement CHIPS act in this area what would that mean for America competitively uh particularly when we start talking about what countries like China are doing. So it's not only the economic opportunities that producing the chips here have for communities across the country. Uh there's a real national security implication.
▶ 3:00:35Yeah. Chips along with data and power and human intelligence fuels this AI revolution we're in the midst of. If you don't have those chips and you're beholden to other countries and other supply chains, uh that's a real vulnerability. Yeah. Which means China rules the world. Very scary. Mr. Chairman, the gentleman yields. The chair recognizes Mr. Bence from Oregon for his five minutes of questioning. Thank you, Mr. chair and thanks to all of you for being here.
▶ 3:01:04Uh is there is there a a reason that we should have a more organized approach, Mr. Wang, to the uh approach that we're using to try to achieve this uh win in artificial intelligence? And before you answer that, tell me what a win is in artificial intelligence. I I know when I was reading the Oenheimer book uh and uh Turing's Cathedral and other such literature, well more Oenheimer, the goal was a bomb. What's our goal in AI?
▶ 3:01:33Uh AI is is as has been mentioned uh a unique technology because it has such uh broadreaching implications. It can be utilized uh to empower our economy and enable our our industries to grow. It can be used for science uh in accelerating scientific discovery, helping us do s do things such as uh you know solving fusion or finding a cure to cancer. Um, and it can also be used uh for as a as a weapon and used in military contexts.
▶ 3:02:00And I know I asked you uh to tell me if our approach is the proper approach and we'll get back to that in a second, but uh as Lincoln said, the way you get things done is to change public opinion. And and the great thing about this hearing today is what we're trying to say is this is an existential issue. This is so important. We need to wave environmental rules. We need to push things aside. We need to create exemptions. We need to get past this hay stack of of uh of uh obstacles that we've created for ourselves to protect things.
▶ 3:02:31But it takes forever now to do anything here and we don't have forever. So what I'm really asking is make the best argument you can to America right this minute about why this is an truly absolutely necessary thing for us to set these other important things aside. And it it has to be more and maybe it can't be but your best argument I'm gonna ask everybody else the same question but go ahead.
▶ 3:02:54If we fall behind the Chinese Communist Party uh this technology will enable the the CCP as well as other authoritarian regimes to utilize the technology to uh over time effectively take over the world. Um you know they'll be able to export their ideologies. They'll be able to utilize as a military technology to invade other countries. um and they'll be able to use it for uh effectively spreading their their regime in a more broad way across the world. And so what's missing of course is you say use it.
▶ 3:03:24The definition of it is is going to become more and more important so people actually can grasp this broad phrase of AI is so general. Your turn. Well, I think it um it's it's really critical that we um not only maintain our leadership in terms of the the algorithms and the the um the data uh structure uh approach to being able to enable the AI applications.
▶ 3:03:51But absolutely the hardware semiconductors, logic, memory, these are it's absolutely critical that we're able to maintain our advantage. They they are critical. But what I'm trying to get at here is public opinion has to understand why they are critical. why it's absolutely essential that we win this race to a goal that's not as clear as I would like. Uh Dr. Schmidt, um in 5 to 10 years, every American citizen will have the equivalent of an Einstein on their phone or in their pocket.
▶ 3:04:20This is an enormous increase in power for humans. What if that Einstein is a Chinese one? And I'm going to shift back to Mr. Ring for just a minute. uh because of frankly your age as compared to those other on the panel. So if if everybody's going to have Einstein uh available, how would you suggest to teachers that they address this in in the classroom?
▶ 3:04:47I think it's important uh frankly I think AI will be an immense opportunity for uh humans and for industries to be able to leverage as a core technology. Our view is that you know in many ways um our role uh you know humans role will go towards supervising and managing these AI systems these AI agents if you will um in a in a and give ourselves frankly more leverage.
▶ 3:05:10So I think the key for teachers and for education system is to teach people how to leverage AI systems, how to use them, you know, how do you embrace the technology as a tool, as something that enables you to do more things, better things, you know, more ambitious things and that would mean that all of our teachers have to understand how to use this new tool. And I'm I'm I'm going to be out of time, but I I was interested, Mr. Turk, in the in the remarks you made about trying to recover and bring back to the United States manufacturing capability.
▶ 3:05:40I know that Micron's the only memory chip maker we have left here. And and so I think the tariff concept is exactly that to try to in some fashion get us back to where we need to be as we as we watched all those different important jobs flee now doing our best to get them back and and the the real question is how to do it. And I'm out of time, but it's certainly incredibly important. Thank thank you all. I yield back. The gentleman yields. Uh the chair now recognizes uh the gentle lady from Massachusetts, Miss Tran.
▶ 3:06:10Thank you, Mr. Chairman. Uh Congress must lead in advancing uh pro- consumer pro- innovation uh AI agenda. Clear guard rails and regulatory certainty will fuel not hinder that innovation. Uh so to understand what that looks like for everyday Americans, let's just start at the beginning of the AI pipeline uh with research and what we need to be doing to set uh the conditions for AI capacity.
▶ 3:06:40Foundational research drove breakthroughs like transistors, the internet, and large language models. Our adversaries get this. Um while federal uh R&D funding is being cut, the Chinese government is scaling its investments. For example, China is outspending us by more than double in fusion energy research and commercialization. Uh Dr. Schmidt, in your testimony, you noted the importance of ramping up fusion energy research.
▶ 3:07:10Commonwealth Fusion uh is in the district that I get to represent. uh the fusion and AI leaders that I regularly speak with tell me how important public private partnerships are for advancing new technology and moving towards commercialization. How important is a strong federal research enterprise for domestic innovation including infusion and an AI? Um, Commonwealth is an example of American exceptionalism.
▶ 3:07:38As you know, their development of these incredibly powerful batteries, sorry, magnets, excuse me, that was done in research at MIT shows you the path. You do it at MIT, you do a spin out. It was done collaborative with MIT with other investors. People have put billions of dollars into Commonwealth, including myself. I'm also the chairman of a competitor company on the West Coast. That's how the American system works.
▶ 3:08:02The current 15% uh indirect cost issue is hurting American science and it needs to be addressed. If there are issues in specific programs, do it surgically. The damage that is being done to American research broadly speaking will harm the country for the next 50 years. This is the time to reverse this. Your answer is exactly why the federal government must bolster and not squander its research capacity.
▶ 3:08:32Whether it's pushing away international researchers or gutting science agencies like NIH or the National Science Foundation, undermining research, the first step in the AI pipeline threatens our ability to win, which I believe we all want to do. Uh I'd like to turn to AI development, which depends on computer chips. Uh, in 2022, Congress passed the Chips and Science Act to bring chip production back home.
▶ 3:08:58China sees the same strategic value and is implementing a massive state sponsored campaign to strengthen its semiconductor supply chain. Mr. Beta, companies like Micron have received billions through the chips act to expand chip factories in the United States. How important is it that the federal government fully implement the chips act to ensure that Micron and other firms are able to bolster their domestic manufacturing capabilities? Thank you, Congresswoman.
▶ 3:09:26And um you're right that um you know our Asian competitors uh do have uh you know large cost gaps cost deltas versus our operations here in the United States 35 to 45% range depending on where in Asia and um those countries uh are also incentivizing their domestic uh uh companies which uh creates competitive disadvantages for the US uh US companies and it's absolutely essential that we are able to um extend and expand the investment and tax
▶ 3:09:56credits that were passed as part of that legislation so that the um the uh spring of of new facilities that has started can continue in bloom over the next decade. Y thank you. Uh you know, President Trump says he wants to revive American uh manufacturing, but he is gutting the chips program office and floating repeal of the chips act alto together and that just doesn't add up. Um, finally on AI deployment, uh, to benefit from AI, people need protection.
▶ 3:10:26AI isn't flawless. It can mislead. It can m make false predictions. Uh, it can expose personal data. Yet, yes, we must beat China. Uh, but we don't need to become China. America must lead with its values, especially privacy. Our tech laws should reflect that. Mr. Wang, in your testimony, you affirm the need for effective AI guard rails. This committee has repeatedly come close to passing a federal privacy standard based on data minimization and transparency.
▶ 3:10:56How important are privacy protections as a guardrail for AI? Uh you know we we strongly support Congress's desire to get data privacy legislation done. Um ultimately, uh what we find critically important is that again I I've mentioned this a few times that we have one federal framework uh so that we don't have a patchwork of of various frameworks throughout the country. Yep. This committee has a lot of work to do. Thank you so much for your testimony.
▶ 3:11:26The gentle lady yields. The chair now recognizes the gentle lady from Iowa, Miss Miller Meeks. Uh thank you very much uh to the chair uh and ranking member for holding this extremely important uh hearing on uh AI energy and global competitiveness. Iowa's first district has become an important contributor to our nation's AI infrastructure. In February, Cedar Rapids announced its largest economic development investment in the city's history.
▶ 3:11:52A 750 million partner uh dollar partnership between the city Alliant Energy and QTS to build a major data center campus. The project will be bring hundreds of construction jobs and high-tech positions while featuring innovative water-free cooling systems that address resource concerns. It's also home to Azour's largest uh supercomputers which Microsoft built for open AI to train breakthrough AI models.
▶ 3:12:17This cutting infrastructure in our state's heartland demonstrates how communities beyond traditional tech hubs can play vital roles in advancing AI innovation. As we examine these technologies, I'm particularly interested in how we ensure reliable power generation for these high demand facilities. Iowa's diverse energy portfolio positions us well, but we need significant additional generation capacity nationwide to meet growing electricity demands for AI, domestic manufacturing, and residential demand.
▶ 3:12:46I'm e eager to explore how we maintain America's energy uh competitive edge, and especially against China's targeted effort to become the global AI leader by 2030. The decisions that we make today about regulation infrastructure will determine whether the United States mains its leadership position and how critical this is. As has been mentioned earlier, Mr.
▶ 3:13:06Wing, I was impressed by MIT's AI innovation when I visited there a few years ago, but concerned to learn about the CCP's whole of government approach to accelerating Chinese AI capabilities. With the recent emergence of models like Deep Seek, how would you characterize our current competitive position against China, specifically in the areas of data? And I think you've answered this partly. Uh computing algorithms and workforce development.
▶ 3:13:32It's an important question and um you know I always I always you know AI really does boil down to to its ingredients and these ingredients are the ones that you referenced computational power uh data algorithms and and ultimately the workforce that we have to support it. Um when it comes to computational power uh we are still ahead as a country but uh we have to we have to be very diligent to ensure that we stay ahead.
▶ 3:13:55Um we're lucky that the leading chips in the world are Nvidia chips um some of the chips from Micron and others um which are uh which are the forefront of the industry and the envy of the world. Uh but we need to maintain those leads and we need to to think deeply about how we do that. Um when it comes to algorithmic uh the algorithms um you know I would actually say we're probably on par at this point with China. They you know we used to have a meaningful lead.
▶ 3:14:18Most of the most innovative algorithms are American innovations but they've been very quickly replicated and at this point it's not clear that we have a lead. Um when it comes to data this is where uh China has an immeasurable lead. They've they've invested in it for years you know nearly a decade of investment into data sets to fuel their AI development. This started with uh their global surveillance programs and when they you know instituted large-scale AI for facial recognition and other technologies throughout the country um and has continued to today.
▶ 3:14:49Uh we need to to figure out as a country how we achieve data dominance uh and how we can do that both in the public sector as well as across the private sector. Um and then lastly on the workforce uh this is this is an important point. we uh as a country uh again the workforce is what fuels every component of this uh of these sets of innovations. So we need to ensure that we as a country are setting up the right programs to empower the AI workforce of tomorrow. Thank you. Mr.
▶ 3:15:16Bonnie, in your testimony, you stated that the US is not on track to keep pace with projected energy demand and that unless the US makes substantial policy shifts, access to affordable and reliable power will begin constraining Americ's manufacturing renaissance. During our hearing with the nation's grid operators last month, they expressed similar concerns. Your testimony specifically highlighted uh the Broadman to Hemingway transmission line project that's faced nearly 20 years of permitting delays.
▶ 3:15:42Can you elaborate on how these permitting challenges directly impact Micron's expansion plans and competitiveness compared to China's ability to rapidly deploy energy infrastructure? Thank you, Congresswoman. The the boardman to Hemingway line is just an example.
▶ 3:15:57It's a project that is um I think 300 miles long and has been on the drawing board for almost 20 years now and it's we were joking earlier that it it's approaching its 21st birthday almost in terms of how when it was proposed until today and still hundreds of millions of dollars spent on permitting. It's a it's a project that does span three different states to be able to to connect transmission in the Pacific Northwest.
▶ 3:16:23And because of those kinds of regulations between the different states as well as federal um uh oversight issues uh and regulations, we've not been able to see it even get started. And that's just one example of I'm sure many many other examples of projects which really are needed to be able to bring the grid resiliency that that others on the panel have talked about and that I've called for as well. Thank you. I have a question for Dr. Schmidt on Fusion, but I'm out of time, so I will submit it for the record if you could please answer, but I hadn't heard Fusion mentioned, so wanted to get that in.
▶ 3:16:53Thank you. I yield back. The gentle lady yields. The chair now recognizes the gentle lady from New York, Mr. Okaziocortez. Thank you, Mr. Chair, and thank you to our witnesses for joining us today and offering uh your testimony. Um, you know, this the crux of what we're here to discuss today is where a lot of where the rubber meets the road when it comes to AI and how this actually manifests in the world and the real problems that we're going to have to square and solve.
▶ 3:17:24Uh, particularly as it comes to energy and energy consumption. U, Mr. Schmidt, Dr. Schmidt, you you've written in the past about the energy consumption of AI. Um you mentioned in u in this article here on project syndicate that quote AI guzzles electricity. A single chat GPT query requires 10 times as much as a conventional web search.
▶ 3:17:51And in your opening statement today, you said something very fascinating and compelling I think about um the actual scale of the energy consumption that we are confronting here when you talked about um gigawatts and nuclear facilities. Could you repeat that uh for me very quickly? Um so some math here is and thank you congresswoman. Um the typical data center sorry the typical nuclear power plant is one gawatt. We have roughly 90 of them.
▶ 3:18:20We're talking about 90 gigawatts in the next three to five years needed in America to maintain this leadership. 90 gawatts for the AI data center for the United States. And and the the the reason I want to emphasize this is one this is in this is insane in terms of a build. Why do we need it? Because we're going from chat, the chat GBT that you know, which is language to language, right, to reasoning systems that do thousands and thousands. What they do is called reinforcement learning. They go back and forth and back and forth.
▶ 3:18:50They're not as efficient as our brains and they discover new things. And so we are and uh and I completely hear you on the on on the scale of the technology that that we're dealing with here. And going back to that 90 gawatt number that's the equivalent of to to what you mentioned here 90 nuclear power plants just that we would be developing or the equivalent of that just for AI data centers alone.
▶ 3:19:18Um and of course we're not talking about building 90 nuclear power plants. We're talking about building that capacity which before us here today to be frank and with the current administration is fossil fuel infrastructure. Of course we've talked about mixed uh energy loads but with the investments and what we are seeing in terms of what is getting defunded and what is getting funded and what is being advocated for. This is largely fossil fuel infrastructure and particularly methane.
▶ 3:19:47methane being 28 times more potent in contributing to uh the climate crisis than even traditional uh CO2. Um but what we're also seeing is that in the administration's moves to massively invest in AI, uh we've also seen the fossil fuel market be tightly associated with this.
▶ 3:20:13In fact, um, the day after Trump announced his $500 billion AI Stargate initiative, gas prices in the market went up 5.3%. And after the deepseek announcement from China, which uh announced that they use consume 50 to 75% less energy, gas prices fell 8%.
▶ 3:20:37And so increasingly we are seeing fossil fuel market speculation seeming to start to intertwine with the development of the AI industry. And this has a problem for working people. And this is the part that we need to square in New York. Con Edison bills that's our kind of local energy provider are up for some families are paying $1,200 a month to pay their energy bill.
▶ 3:21:03And we're here talking about massive investments, not to lower their bills, but for ultimately infrastructure that's privately owned. Mr. Turk, if a utility invests in a new substation so that gas generation for an AI data center can connect to the grid, will that utility typically pass those costs on via its electrical rates? Well, I think you've hit the nail on the head here, right? We don't just need new electrons for AI.
▶ 3:21:32we need them for consumers, right? And we need to have downward pressure on prices, not the opposite. And so that's what we need to keep our eye on the ball, including and especially with the IRA tax credits. What we're talking about is average households paying $200 more per year if those tax credits are repealed. For citizens in New York, it's $400 per year more.
▶ 3:21:51And so I think so but at the core of the question without you know if we currently go on this path the increases in that energy consumption from AI get passed on via bills. That's exactly right. It's a competitive environment. Uh we have increasing demand uh if we don't have a range of resources especially solar and storage which are the cheapest uh resources to bring on quickly right now in our country.
▶ 3:22:17If we increase the prices of that, everyone's going to feel it and consumers in fossil fuel prices are certainly more volatile than we That's right. Thank you very much. The gentle lady yields the chair now recognizes the gentle lady from Florida, Miss Lee. Thank you, Mr. Chairman. Mr.
▶ 3:22:34Wang, I would like to go back to your testimony of earlier today specifically toward the end where one of the things that you touched on was how we as Congress might empower and utilize NIST to help us in our promotion of uh artificial intelligence. One of the things that you mention is that uh NIST needs more resources uh to be able to complete relevant measurement science such as standards and frameworks.
▶ 3:23:00Would you tell us a little bit more about elaborate on those standards and frameworks and what you think NIS could be doing that would be constructive? Ultimately, as AI develops as a technology, it's very important that we have uh what we what we call test and evaluation regimes that we're able to both test and evaluate the performance of these AI systems, understand their limitations, as well as do as as as um other of the panelists have mentioned, do extensive red teaming on these AI systems, understand how an adversary would be able to utilize AI or hack
▶ 3:23:30into our AI systems um to uh to harm us. Um, you know, this this work is incredibly important and serves as a foundation that we can use to export American AI standards globally. And this is, you know, this is really the strategic uh move for America, which is how do we ensure that the way that we think about AI both embedded with our values and our democratic values as well as how we think AI should um should be developed globally is as exported as broadly as possible throughout the world.
▶ 3:24:00Um you know we saw I think in the uh in the last few generations of technology um the Chinese Communist Party actually be quite strategic on this. The belt and road initiatives um their use of Huawei technology for 5G you know they've in many recent uh developments major developments in advanced technology they've focused on exporting their technology and making sure that Chinese technology is the is the global standard. we need to do the opposite with AI.
▶ 3:24:25And the the beauty of uh the situation that we're currently in is that uh many many countries, you know, Japan, France, the UK, um uh India have all established uh AI safety institutes that are all looking towards the testing that we're doing in the United States and the standards that we're enforcing in the United States for them to institute institute their own If we are able to develop and then effectively export that measurement science, would you elaborate on how it is that you think
▶ 3:24:55that will help promote democratic values? And similarly, if we fail to do so, what do you anticipate that we will see uh if if we do not create those standards and share them globally? Ultimately um you know as just as a simple example um let's say that we uh we institute as part of our as part of our test and evaluation systems um certain guard rails around factuality.
▶ 3:25:24So the the AI systems um uh you know or certain guard rails around uh you know whether or not the AI could be used to to create bioweapons or whatnot that would totally eliminate certain classes of risks of a CCP model being used uh globally to to you know uh perpetuate their ideologies or perpetuate um uh you know perpetuate instability globally.
▶ 3:25:48You know there's there's we have an immense ability to ensure that the United that the American view of AI which is a democratic technology that can be utilized by the people for the people to ultimately empower industries that that is how the entire world views the technology and and it's a it's a it's a fixed window of opportunity. We will not have this opportunity forever. At some point all of the other countries will will uh will start instituting their own AI standards. And so we need to act quickly.
▶ 3:26:18One of the things that you mentioned is your assessment that NIST would benefit from having additional resources from Congress in order to be able to undertake this activity. Uh do you have a perspective on how that looks? Whether it is dollars, whether it is people, if there's a certain type of workforce they require. Do you have any perspective on how we could better equip NIS to be ready to do this? Uh yes, I think I think all of the above are important.
▶ 3:26:45I think ensuring that they have the dollars, ensuring that they have the headcount and one of the things that I think is very critical is that they are able to bring in uh and leverage cutting edge AI talent as a part of NIST to help define these standards globally because these are very advanced technical questions that need to be answered but ones that will have immense benefit to America and our economy long to the future if we succeed. Thank you, Mr. Chairman. I yield back. The gentle lady yields.
▶ 3:27:12The chair now recognizes the gentleman from Massachusetts, Mr. Aenclauss. Uh, thank you, Chairman. Dr. Schmidt, it's good to see you again. You had come and spoken to the Select Committee on China, and you were elucidating then, and I've enjoyed hearing your testimony today as well. I I was hoping you could tell the committee a little bit about a a famous Google paper in 2017 called Attention Is All You Need.
▶ 3:27:33Now, you were no longer executive chairman at that point, but you had been stewarding the company for the 15 years before that and I'm sure is well aware of how that publication came to be. Can you give us like a minute backstory? Um, I was in fact still executive chairman and the interesting thing about that paper is when it came out I didn't even notice it. Shows you asleep at the wheel or something. The six authors all became hugely famous because they came up with a way of building scalable intelligence.
▶ 3:28:03Before that the RNN and CNN not not media CNN convolutional neural network architectures were slow and the attention is all you need allowed you to essentially devolve the computation into subdividable things which could scale the transformer paper and the T in GPT is transformer is the underlying architecture that has enabled this explosion.
▶ 3:28:28Am I right that that transformer architecture in the 2010s was really came became coherent around the problem of natural language translation? Um not really. The transformer architecture was essentially a refactoring of the technologies of the time into a more scalable architecture. uh specifically that you could have federated computing. You could have lots of different computers doing things at the same time. It's the easiest way to explain it. And it was a real breakthrough.
▶ 3:28:59They'll ultimately win the equivalent of Nobel prizes for it. Well, I was I was looking I was doing some research about the what's been called the Transformer 8, the eight authors of that publication. And they're almost like the PayPal mafia of AI. I mean, what they've gone on to do is remarkable. Here, what here's what else is remarkable. Of those eight, seven are immigrants. Yeah. And the eighth is the grandson of refugees who came to the United States fleeing persecution.
▶ 3:29:26In fact, twothirds of top AI startups are founded by immigrants and most PhD level AI talent in the United States is foreign born. Dr. Schmid, can you describe the impact of immigration on AI, America's AI competitiveness? I was in a conversation last week in London where people were talking about people leaving the United States AI companies to move to London because they couldn't work here anymore. That is insane. That it's so counter to American national security.
▶ 3:29:56It's like crazy. Um from my perspective, the most important thing American can do is look for high skills immigration. These to describe how hard this stuff is. These are PhDs in math. I have no idea what they're doing. And they're inventing these incredible algorithms. We need all of them in America. Every single one of them. Physics, chemistry, you name it. We need them all. And yet the Trump administration is currently eroding due process for immigrants in this country, whether they have green cards or student visas. They are deporting students.
▶ 3:30:26They are creating a climate of fear and anxiety on some of our best campuses. Go ahead, sir. It's it's actually worse. We we agree. It's actually worse. Uh people are people are being thrown out of the universities that are doing AI research. Universities have shut down their hiring pipeline uh and they need AI professors and and the people will otherwise go to industry. So the damage being done to the universities is really really profound.
▶ 3:30:50It's very very important that we understand that American leadership in the in research which you understand very well from where you are is the cornerstone of our future. We will not get there. Meanwhile, China is pouring an enormous amount of money into the same groups. Do you think if the Trump posture towards immigrants, student immigrants in particular, universities more broadly, if that persists, can America beat China and AI? No.
▶ 3:31:16In fact, uh when I was you all appointed me to be the chairman of the National Security Committee on Commission on AI and we looked at this very carefully. What was interesting is that Chineseborn um contributors were often part of the key papers. They were not the lead offer, but they were part of it. If you'd get rid of those people, and in particular, they go to China, right? The leadership literally moves. I'd much rather have them be here. And people say, "Well, you know, they're are criminals." They're not criminals. They want to be in the United States.
▶ 3:31:46If they are criminals, arrest them. They're Americans by choice. Uh, not only is Donald Trump providing a massive opening for China with his xenophobic immigration policy, he's also providing that massive opening for China with his trade wars that's bringing Europe and China closer together. Uh, with that, I yield back, Mr. Chairman. The gentleman yields. The chair now recognizes the gentleman from California, Mr. Oberoni. Oh, thank you very much, Mr.
▶ 3:32:11Chairman, and I'd like to thank Chairman Guthrie and the Energy and Commerce Committee for having this hearing on a topic that's very close to my heart and something I think is of immense national consequence to our economy and our country. Um, Mr. Wang, it's great to see you again. Um, in your testimony, you were talking about the steps that must be taken to ensure US continued leadership in AI.
▶ 3:32:35And uh I was very thankful that you had some very specific asks of Congress and the administration and one of those was that we adopt a regulatory framework that's sector specific and use casebased and uh wanted to ask you uh could you elaborate a little bit on uh on what you mean by that and uh and how we would go about enacting it? Yeah, so ultimately um what we need as a country is to ensure that from a technology development standpoint that we do not slow down.
▶ 3:33:05We need to ensure that AI as a technology moves forward as quickly as possible and that includes embracing the technology and ensuring that we have we have the we have room to innovate but the application of that technology towards certain sectors or certain specific use cases um in the economy are areas where I think um you know there probably needs to be some level of regulation or at least some level of of of guard rails in place.
▶ 3:33:30You know these could be industries like um the medical industry, the pharmaceutical pharmaceutical industry, the financial services industry and others um you know industries that already have some degree of regulation to protect uh consumers and protect Americans.
▶ 3:33:43you know we can in many cases we can we can utilize those same provisions or those same regulations and then there might be some cases where there are some gaps right uh the uh artificial intelligence task force in the house issued a report in December that made exactly that same uh recommendation and uh I think the the exact finding was uh that we regulate tools not outcomes not tools and AI is a very powerful tool but is a tool if we focus sorry,
▶ 3:34:13regulation on outcomes, then we can capture all the different uses of the tool. Um, you also talked about the need for a single federal standard for regulation and Congressman Dunn was on the way to to asking you about that and unfortunately ran out of time. So, I wanted to give you a little bit of space to explain what you meant by that. Yeah.
▶ 3:34:32So um you know as a as an AI company and I think I think what we ultimately want as a country is to ensure that uh that our industry can continue developing advanced AI systems and continue um driving American leadership. Uh you know the worst case scenario for us is actually that there are 50 different that every state adopts a different regulatory standard and we have to uh you know operationally comply with 50 different regulatory standards.
▶ 3:34:58I mean it quickly becomes impossible especially as you consider you know in in a lot of cases the the the way that we develop AI is we develop you know one large model and then we start applying that model in all sorts of different industries and use cases um and jurisdictions.
▶ 3:35:13And so we need as as an industry and as a country um one clear federal standard uh whatever it may be but we need one we need clarity as to one federal standard and have preeemption to prevent this outcome where you have 50 different 50 different standards.
▶ 3:35:29Just to put a finer point on this, um, you know, we do not want our our American companies spending all their time figuring out how to comply with every state standards, whereas the the Chinese models and the Chinese companies will just race ahead on on innovation, right? Uh, that's another conclusion that we completely agree with you. The task force report, we had a whole chapter on this issue.
▶ 3:35:50And uh let me just point out the fact that since then, just in the last couple of months, we have uh at last count 958 bills pending in state legislatures across the country on the topic of AI regulation. And I'm sure it's going to grow to be several thousand just in this year. Uh if we allow this regulatory landscape that complicated to exist, I actually think that scale is probably well suited to that because you've got the legal sophistication to deal with that.
▶ 3:36:18But who does not have that sophistication are two people at Caltech, see what I did there, not MIT, Caltech, go trying to start the next uh scale. So, uh I think we definitely we have like we have a limited amount of legislative runway to be able to get that problem solved before the states get too far ahead. Um and one last question for you, Mr. Wang. You had mentioned the need to establish a national AI data reserve. Could you talk a little bit about why that's so important?
▶ 3:36:46if we you know ultimately national security is the responsibility of of the government um and and uh our government's data particularly our DoD's data and our and our data relating to national security is so vital and valuable to ensuring that our AI systems are able to defend our country defend our men and women and ultimately um ensure national security broadly speaking.
▶ 3:37:10So the the necessity of the national AI data reserve is so that you know in in 10 year 5 to 10 years we're not sitting here seeing how advanced the Chinese systems for defense and intelligence and you know cyber cyber warfare and other systems are because they have an integrated data approach versus our systems which would be dramatically behind. Well, thank you very much for your testimony. Sorry I didn't get to the other witnesses. I have a million questions. We'll submit that for the record. I yield back. The gentleman yields.
▶ 3:37:39The chair now recognizes the gentleman from Louisiana, Mr. Carter. Thank you, Mr. Chairman. Um, and thank you to our witnesses for joining us today. I'm glad we're discussing the need to upgrade our infrastructure for the 21st century economy and provide for all Americans access to cutting edge technologies championed by our witnesses here.
▶ 3:38:00In January, Louisiana became the first state to secure federal approved for deploying $1.3 billion dollars in broadband equity access and deployment uh otherwise known as bead funding. This achievement highlights the bipartisan nature of Louisiana's commitment to universal connectivity and to set standards for states regarding broadband access.
▶ 3:38:24The state's bead bead rollout plan began under the Democratic Governor John Bell Edwards and was completed under Republican Governor Jeff Landry, who called it a generational investment that will create thousands of jobs, drive billions of dollars in economic growth, and transform Louisiana's communities in all 64 parishes.
▶ 3:38:46The state's plan will connect approximately 140,000 locations to high-spe speed internet through funding awarded to 20 internet service providers with nearly 70% of the funds awarded to Louisiana companies. More than 90,000 of these locations were set to transition from zero connectivity to futureproof broadband fiber.
▶ 3:39:10Although these broadband investments will drive significant economic growth for the state, creating approximately 10,000 new jobs and generating an estimated 2 to3 billion dollars in new revenue for Louisiana companies.
▶ 3:39:25However, since the Trump administration took office, just a week after Louisiana received approval, its final approval to move forward on its proposal, the Commerce Department has withheld final funding to their approval that would have otherwise had shovels in the ground installing high-speed broadband infrastructure today. Not aspirational, but now. The unexpected delay has stalled progress.
▶ 3:39:54frozen investments made by small internet service providers and contractors and left rural communities still waiting on the promise of broadband access. Just recently, Meta announced that they were building a roughly $10 billion data center in rural Richland Parish in Louisiana, an area that would have benefited from the state's broadband rollout.
▶ 3:40:16In fact, over 600 households within a 10 mile s circumference of the new meta facility would be connected via bead. We also expect that around the data center to grow as facility brings in hundreds of workers including skilled technical The delays around bead roll out mean that these workers for the10 billion advanced data center may lack high-speed broadband at home, threatening yet another
▶ 3:40:47huge investment in my home state. The freeze and bead funds yet another example of how the Trump administration has shown chaos and uncertainty for businesses trying to make major investments in technology and energy on top of the past week of economic turmoil and worldwide market crashes. This is unacceptable. Mr. is it Bajita? It's close enough. Enough.
▶ 3:41:13Um, how important is quality of life for your workers when you are looking to grow your operations in new areas? Would considering making major investments in the area where your workers are and their families lack access to the internet be a major Absolutely.
▶ 3:41:32We would like to ensure that we have um a workforce that is uh highly skilled, highly trained and can um uh and you know all the jobs that we're creating with our projects uh you know 11,000 direct jobs at Micron 80,000 direct and indirect jobs those all should be highpaying jobs which will allow people to have a high high standard of uh living and we think that's an important element to ensure our our uh technology leadership as well as our manufacturing efficiency. Thank you m Mr.
▶ 3:42:01Wang, in your testimony, you recommended that the federal government put policies in place to let the AI workforce thrive in America. Would you agree that we're holding back our future workforce by allowing children to grow up in an America without access to high-speed broadband internet?
▶ 3:42:21Uh I certainly think that the ability for our future uh for for for ch our children and future workforce to embrace AI technologies and other technology is going to be absolutely critical to uh you know the future development of our country. So ultimately yes I think we need to ensure that Thank you. I've got four seconds real quickly. Mr. Turk uh our American grid is now facing an unprecedented surge in electrical electricity demand.
▶ 3:42:47How has the Trump administration's blanket refusal to permit large-scale offshore wind projects impacted our country's ability to meet this new demand? So, it's another tool in the tool belt. Why take it off? It's incredibly important along with other sources of power. And I think your point more broadly about infrastructure funding, you need predictability and you need certainty. You don't need chaos. And that's what Thank you very much. My my time is has ended. I yield back. Thank you. Uh the gentleman yields.
▶ 3:43:14The chair now recognizes the gentle lady from North Dakota, Miss Fedorek. Good afternoon everyone. Thank you for being here. It's been an interesting hearing. Dr. Schmidt, you said you think the AI the importance of AI and the challenges we face has been underhyped. I agree with you. I also think that the challenges that our electric grid in this country face have also been underhyped.
▶ 3:43:42Uh the truth of the matter is we are underpowered today and that doesn't even take into consideration the demands that the AI industry brings or the the need and the urgency for us to meet that demand. Um so knowing that uh would you all agree that one of the first things we should be doing uh is stopping retiring of existing resources that are connected to the grid. And I'll just go down the line. M.
▶ 3:44:12Mr. Turk, it real quickly. I don't need a one minute answer. Yes or no? We should stop retiring existing resources if they're still somewhat economic. Yes. But we do need to keep an eye on other goals including climate and we need to make sure we Thank you, Dr. Schmidt. Uh, yes. Okay, Mr. Wang. Yes, Mr. All of the above. Excellent. Thank you. Um uh I think it's not an yes, but it is yes, we need to stop retiring.
▶ 3:44:42This is an urgent need. It everyone has said it's a national security issue. All resources take time to get on the grid. And so when we don't even have enough to meet demand today, then we most certainly and we have growing demand, we most certainly should all be able to agree in a bipartisan manner that we should keep whatever we can right now and then go from there because technologies evolve and they will continue to evolve. Um Mr.
▶ 3:45:09Turk, you had said earlier that you think that you'd said that um solar and wind are the cheapest resources to bring onto the grid. Uh can you elaborate what what do you include in that calculation? Yeah. So I look not only at the levelized costs, but I look at what's actually being brought into our grid right now driven by economics.
▶ 3:45:30And so 93% our independent energy information administration is saying 93% of the new power brought on this year will be solar and storage and wind. Okay. But that doesn't necessarily mean it's coming on because it's the cheapest. Does your calculation include the cost of transmission to bring that online? Well, this is why we need to have and I know you're an expert in this and thank you for your leadership in Neuk in particular with your previous job. We need to have the whole grid. We need to be thinking about uh reconductoring. We need to be thinking about great enhancing technologies.
▶ 3:46:01We need to be thinking about transmission too. We need about think about it holistically and systemically. Exactly. And I support Gats 100%. It's not the 100% solution though and it's it should not be overstated because I think a lot of people who don't understand this hear things like that and think there think there are simple solutions when really they're far more complicated.
▶ 3:46:23And the truth of the matter is when you consider solar and wind as being the cheapest, the cost of the transmission is not included in that calculation. Nor is the cost of all the backup generation that's needed to provide power when solar and wind aren't available. Those have to be included in our calculations when we're talking about costs because the people who pay for that, they notice that those aren't the cheapest things because it's all included in their bill.
▶ 3:46:53Nobody else soaks up those costs but the final customers who pay the bill. Um I'd like to ask one more question of all of you. So I think that in an urgent time like this, it's more important than ever that the signals that this federal government sends through its policies provide clear messages and clear instruction about what we need the most. We had all the grid operators here a week ago to the to the person.
▶ 3:47:22They all said what they need now is dispatchable power. Knowing that, is it reasonable for the federal government to continue to incentivize resources that are not dispatchable? And I'll start down here at the end. Should we be sending that signal? If what we need is dispatchable, why are we sending strong signals that you should bring on things that aren't dispatchable through tax policy? I think that, you know, I mentioned all the above earlier.
▶ 3:47:49I think that we need to think about technologies that can uh and investing in technologies that will be able to contribute longer term. We shouldn't take away from that. I mentioned um in my prepared remarks um you know some nuclear technology that we stopped investing in that you know probably looks today to be shortsighted. Um but at the same time we need to be uh focusing on the technologies on the uh the sources of energy that can um uh support the uh the demand today. Thank you Mr. White.
▶ 3:48:16Uh I'm not an energy expert, an AI expert, so uh only know the best answer this. Okay, Dr. Schmidt. Um if you take all of the subsidies away of oil and gas and all the ones around renewables, you get a different calculation. Given we have the oil and gas subsidies, it's I think it's fine to have the renewable subsidies. Yeah. Key thing is solve the storage problem, which I think has largely been solved. That creates dis dispatchability. Thank you. I yield back. The gentle lady yields back. The chair recognizes the gentleman from New Jersey, Mr. Menendez, for five minutes.
▶ 3:48:47Thank you, Chairman Guthrie, uh, for holding this hearing this morning uh, afternoon, I guess. Now, Dr. Schmidt, in your testimony, you state that securing America's energy future requires bold strategic federal action and investment. One example highlighted by both you and Mr. Turk is the potential for fusion energy, which is generally supported by both Democrats and Republicans. Dr. Schmidt, can you briefly describe the potential fusion has for the future of our domestic energy production? Um fusion is different from fision. Um it's a very different process.
▶ 3:49:17It's the it's the technology that's inside our sun. There are two main approaches. One is essentially it's called a tokamac. You essentially create a plasma that floats. The plasma is so hot you have to control it using magnets and AI to hold it otherwise the walls would melt. There are a number of companies in America that are using that approach. There's an alternative approach which is a pulse fusion.
▶ 3:49:40This was funded initially through something called N in Liverour way back when and it looks like the pulse pulse and what you do is you create a magnetic field which causes a collapse that causes electricity and the electricity generated is greater than the electricity to cause the pulse. It's called Q greater than one.
▶ 3:49:58The timeline of these things is demonstration for a number of these companies by roughly If you make some assumptions about the number of electricians and the scale of the problem and the the and the devices are typically 400 megawatts. So think of the number of 400 megawatt sort of power sources and you sort of take the current power source coal power nucle uh basically natural gas whatever and you put this fusion thing in it. That's the model.
▶ 3:50:26The problem is when I look at the time frame, you're not until 2040 to 2045 when you have abundant fusion, right, to get onto the green and make it part of our daily. But having said that, this is an this is an area where America will lead. It should be a source of great pride for America to lead in this for the world. I agree with you. And how important is federal funding specifically for the US National Laboratories program to advancing new technologies like the DOE work in this is fundamental.
▶ 3:50:55And that's true of the labs and all of the stuff I'm talking about, the people that I've hired in my company are all coming out of the labs. Thank God. And thank you for that. And so just yes or no. If this program were to see its funding cut or significantly reduced, would that hinder our ability to harness this new technology? It would be horrific. We need much more funding in these areas. See, I agree with you.
▶ 3:51:13But last month at a space science and technology committee hearing, leaders from the Department of Energy sound the alarms about tens of millions of dollars that are crucial to research development being put on hold because of President Trump's funding freezes across the federal government. Um, Dr. Schmid, your testimony, you mentioned the need to dramatically increase funding for energy sector cyber security. Dr. Schmidt, again, just yes or no. Should the federal government take the lead on having a strategy to combat cyber attacks to our critical infrastructure? It it has to. Yeah, I agree.
▶ 3:51:42But President Trump recently signed an executive order that put states and municipalities at the forefront of our nation's cyber attack response process instead of the federal government, weakening federal in investment and disaster preparedness and creating a patchwork plan for attacks to our critical infrastructure across the country. Dr. Schmidt, yes or no? Does that seem like a wise strategy? It it's it's it's not a good idea. Remember that we have an incredible cyber force in America under the Pentagon and the National Security Agency.
▶ 3:52:12I do a lot of military work. They are phenomenal. I agree with you and their work should be celebrated and it should sit at the federal government, not state municipalities. I'm a complete agreement with you, Mr. Wang. In your testimony, you call for the establishment of a national AI data reserve. Your testimony also notes that the right regulatory framework maximizes innovation while still creating proper guard rails. Mr. Wang, yes or no? Should guardrails be placed on the government's collection of sensitive data? Yes. Yeah, I agree.
▶ 3:52:42Um, but here's the thing, right? So, the Trump administration is currently weaponizing data that they have within their control, including family sensitive personal information that's collected by HUD and IRS to target immigrants, mixed status families. Right? So, I agree that that having the data is the power, right, that we will be able to use in terms of AI, right? and the federal government having a reserve or a collection of data is how we fully harness AI, right?
▶ 3:53:09But this administration's undermining our belief and trust in the federal government's ability to properly hold data and not use it and weaponize it, which this administration is. This is this is my challenge real this is my challenge with Republicans right now is that they are seeing all this stuff happen in real time. Right, Dr. Schmidt? talked about an all the- above approach to energy production, but they want to roll back investments in renewable energy and they sit here every week and make it seem like it's business as usual.
▶ 3:53:38You are their witnesses and you're telling them we need to reverse course on what this administration is doing and they remain silent week after week and that is the challenge. And by the way, people have gone over on the other side. I'm 3 seconds over Mr. Carter. 3 seconds. Okay. But this is something you all need to be accountable to the American times and the gentleman from Georgia is recognized. Thank you, Chairman Guthrie. Thank you, Mr. Chairman. Thank you all for being here. And thank you, Mr. Chairman, for holding this very important meeting.
▶ 3:54:08Our artificial intelligence is transforming every aspect of our economy and our society. As we well know, from energy and communications to national security and healthcare, AI is both ex presents opportunities. I am u very interested in health care and chair of the health subcommittee. So I want to give you an example. Health flow.
▶ 3:54:30Health flow is a company that is applying artificial intelligence to transform the diagnosis and treatment of coronary or artery disease which kills one in five Americans. This is significant. Using a standard CT scan of the heart, heart flows algorithms can determine blood pressure and flow in the coronary arteries, allowing physicians to determine the severity of disease and whether invasive treatment is needed.
▶ 3:54:54In fact, heart flows technology has proven to decrease the rate of heart attacks and save the Medicare program more than $3,100 per patient. Per patient. Our job as lawmakers is to make sure the US continues to lead in AI innovation while protecting American values like data data privacy, reliable infrastructure, and fair competition. Mr. Smith, Dr. Smith, I want to ask you, startups play a crucial role. We all know that they play a crucial role in driving innovation in technology ecosystem.
▶ 3:55:25How can we avoid creating regulatory structures that only large companies with extensive legal teams and lobbying power can navigate? Um I agree with the premise of your question sir. Um the innovation that's occurring in startups is phenomenal. You see completely new techniques using AI. A typical example would be cancer scoring right where you have a bunch of things.
▶ 3:55:49I'm part of the Mayo Clinic uh board and so forth and they have they're spinning out startups to do precisely this so it can be done. Um we need to have the entire ecosystem of venture capital and so forth behind the image that you described. Exactly. And not just where the bigger companies are the ones who are doing this. And may I add may I add that some of that is actually the data problem that that that Mr. Wang keeps talking about. um many of the pro the many of the startups cannot get the data that they need for various regulatory reasons.
▶ 3:56:19A t simple example would be that if you had opt opt out of privacy things for healthcare um that people could for research that you could have research pools then you could accelerate that. There's a whole bunch of approaches there that are reasonable trade-offs. Okay, le let's talk about the role that AI is going to play in developing new treatments and cures and we know that's going to be the case.
▶ 3:56:40How how should lawmakers be thinking about integrating AI tools into HHS and CMS and FDA to create a more efficient process like quicker drug approvals. One of the well the biggest problem with drugs is the phase three trial cost and the time exactly uh I'm involved with a startup that has a new approach using AI to simplify that. We'll see if my startup is successful or not. Um the current model is static and unchanging. It's not informed by data.
▶ 3:57:09A simple regulatory change to allow better analytics around how you prove that the thing is phase three trial would rel would deliver drug in years ahead of time and years is lives ahead of time and and and we all understand this is this is could be a great benefit. I mean this this could be game a game changer with with diagnosing with making sure that we're doing the right treatments.
▶ 3:57:37AI in in health care is is going to be phenomenal. I'm I'm I I'm very optimistic about that, but it's also going to have some some downfalls and and some things that are dangerous that we need to need to really guard against. But we've heard a lot of promise about how it can cut cost and how it can increase efficiency within the federal government, especially in some of the organizations like HHS.
▶ 3:58:00How should regulators think about cont contracting with innovators to integrate AI into the regulatory and oversight functions that we have particularly in Congress? I'll give you a personal a personal answer. The federal government does a terrible job of procuring software. The federal government does quite a good job of building buying hardware. Software is not managed the same way that you manage hardware. Uh software is never done. It requires constant attention. the teams are constantly turning over.
▶ 3:58:31Instead, the federal government purchases specific contracts for specific outcomes with specific teams. Doesn't work in software. In order to achieve your vision, you have to attack the software problem. The reason our government is so incredibly inefficient in my view is because it doesn't use software It doesn't use software correctly. Have you got an example of that? Uh everywhere you look. Uh I mean if you look at what the tech companies do in terms of integrated software there's no analog.
▶ 3:58:59Every aspect of data in the federal government is insecure. Um all of them are being attacked by the Chinese and others. The systems are so bad that people have to add layers on top to fix them. Many of the underlying databases my time's up. Thank you. And I yield back. So Mr. So Dr. I know you had a a hard stop. Can we do one more? Yes, of course. and we will we have just a handful left, but whenever you're let me know when you need to be excused. I appreciate that. These are very important, but I want to make sure that everybody gets a chance to ask questions. So, Dr.
▶ 3:59:29uh Mr. Mullen, you're now recognized from California. Mr. Mullen, you're recognized. Thank you, Mr. Chair. Thank you to all of our witnesses for your testimony. Uh we've heard from our panel today uh that to compete on AI, we're going to need a lot more stuff, more energy, more materials, more investment, uh more of everything. Uh but steel, aluminum, and everything else that goes into powering data centers costs money, and we cannot win the global race on AI if American businesses can't afford the raw materials to build that infrastructure.
▶ 3:59:59Amid this uncertainty, the majority is considering a repeal of the IRA and the infrastructure law, two landmark laws that have already leveraged hundreds of billions of dollars of private sector investment in our country's energy infrastructure. There are also reports coming out that DOE is planning to unilaterally cancel billions of dollars in grants for hydrogen hubs and longduration energy storage projects that have already received congressionally approved funding.
▶ 4:00:23Rolling back these laws and unlawfully cutting uh committed funding will severely undermine the trust in the federal government that stakeholders have until now at least taken for granted. So uh Mr. Turk, in your time as the deputy secretary at DOE, you interacted with stakeholders across the energy and AI sectors. What will be the worst impacts of all of this economic and policy uncertainty, including uh the tariffs, which were referenced multiple times today on on the investments that are underpinning AI.
▶ 4:00:53So, it's the grants, it's the loans, uh, and it's the tax incentives and, uh, getting rid of, um, or even just causing confusion about whether the grants are actually coming. And I should say on the grants, uh, this was money that you all have already given and this is money already obligated in some instances. And so, uh, the private sector needs to rely on the government doing what it's supposed to do, doing it professionally, doing it without any political, uh, interference.
▶ 4:01:22So I think what it does is it not only uh puts those immediate projects at risks uh but it puts the credibility of the government at risk as well. And if we're going to be successful competing on AI, building out our infrastructure, doing all the other things that we need to do, we need to have credibility in the government working in partnership with the private So uh thank you for that. And I I I fear there will be serious repercussions for our energy system if cuts are made to the IRA uh programs.
▶ 4:01:50are essential uh for energy as energy demand increases as part of AI. But as important as the AI race is, we also have to talk about rising costs. People are paying more, not only at the grocery store, but losing money in their retirement savings. But recent estimates show these tariffs are going to cost everyday Americans an additional $3,800 a year on their utility bills to meet both the AI challenge and cost challenge. It's clear that we need more energy resources and we need to get them online as soon as possible.
▶ 4:02:18Earlier today, you mentioned that um renewables are the cheapest, quickest sources to deploy when it comes to energy. So, what uh Mr. Chirp, what does Congress need to do to unlock this development and ensure that consumers are not hit with the higher costs uh yet again by the Trump administration. So, the good news is you all have done your jobs. Now, we could use more, but you've got the tax incentives, the grants, the loans in place.
▶ 4:02:44Uh what's at risk here is if those are repealed, just two provisions, the investment production tax credit, technology neutral tax credit. If that's repealed, Americans households are paying on average $220 more per year. Just with those two provisions repealed, let alone the other provisions and grants and loans not going out than the way they are.
▶ 4:03:08So, uh, this is the worst way to, uh, keep downward pressure on prices is to repeal these incredibly important tax incentives. Great. Thank you for that, sir. And thank you all. With that, I yield back. Thank you. The gentleman yields back. Your chair recognizes Mr. Griffith. Oh, I'm sorry, Mr. Fry. I apologize. Thank you, Mr. Chairman. South Carolina is experiencing a u spat of economic growth. uh from the grand strand of the PD.
▶ 4:03:36New businesses are opening, manufacturers are investing uh and families are moving in. That growth is a tremendous opportunity, but it does pose some some significant challenges. Yesterday, President Trump issued executive orders declaring a national energy emergency and directing swift swift action to boost grid reliability and cut red tape for energy projects. These steps are both timely and necessary. Power demand is rising sharply.
▶ 4:04:02Uh there's a lot of contributing factors to that, but it is and in South Carolina, nuclear power provides more than half of our electricity, giving us a pretty strong foundation. But permitting delays, premature plant retirements, and transmission bottlenecks threaten not only our state, but all 50 states. Uh we need a federal policy that keeps pace with innovation. That means faster permitting, support for fuel secure generation, and a strong, reliable grid. Um I appreciate the testimony of all the witnesses today.
▶ 4:04:31Uh my initial questions, Mr. Batia, um appreciate your comments on the need to reshore semiconductor chip manufacturing and secure our supply chains in this country. As you noted, China controls an overwhelming majority of global capacity for critical material refining and processing and unacceptable strategic vulnerability on our part. Uh during our hearing with the regional grid operators, we heard uh that regions like New England, as an example, are facing real constraints on natural gas capacity.
▶ 4:05:00that bottleneck uh is is holding back the type of energyintensive investments that we need to support AI and manufacturing. So if we're going to plan for the future where we uh reshore significant portions of our supply chain, how import how important do you believe permitting reform is uh to infrastructure like natural gas pipelines and the like? I I think it's critical and I think um the cost of inaction that we've had over the last several years and continue to have is very very high.
▶ 4:05:30I mean you've heard u multiple data points and testimony around the um the sharp spike in in demand that is uh forecasted both because of the data centers are going to be built as well as the manufacturing in semiconductors as well as other uh other industry segments.
▶ 4:05:47And so um you know after having you know many many years where uh supply and demand has been matched and stable this spike threatens to create a dislocation that could ultimately threaten the viability of some of these projects longer term whether those are um in uh in the data center segment or in manufacturing.
▶ 4:06:04And I think streamlining and um working to be able to remove, you know, duplicative processes between federal and state is something that both both parties can get behind and and states, red states and blue states both can get behind trying to ensure that there is um a streamlined process for uh critical projects to move forward. Sir, do you think that we can realistically meet our energy demands without those simple reforms that you talked about? You know, I'm not sure. I I believe that it we we I don't think we should try and figure that out.
▶ 4:06:34I think we should make sure we move forward with the with the permitting and I think the permitting needs to be across transmission. It needs to be across generation and it needs to be across the all of the above uh sources of energy um investments we need to make. Thank you, Dr. Smith. Appreciate you staying a little bit. Um your testimony laid out the strategic importance of AI in the race with China in pretty stark terms. You mentioned that AI data centers could require up to 10 gawatts of power each uh and that we risk falling behind.
▶ 4:07:01Given what we're seeing across the country though, especially in states with businessfriendly environments, can you speak to the importance of permitting reform and how it relates to our competitiveness in the AI space? When you look at people who have the money, they still can't get the permits and in particular the interconnection permits that are needed to get into the grid. Um, you can solve that problem by, for example, building your own power plant next to your own data center, but that's not particularly efficient. Um, there are all sorts of other issues.
▶ 4:07:30If you look at the cost of, for example, building, I'll give you an example. TSMC built a semiconductor plant in Arizona. Um, and by the time they were done, it cost four times more than in Taiwan. Some of that is labor, some of that is permitting, some of it is government. We are not competitive globally against our key partners and competitors with respect to costs and timing. Thank you for that. And you also mentioned the potential for AI to help manage and secure our grid.
▶ 4:07:58Uh what role do you see for the federal government in AI enabled grid modernization particularly for regions like mine in the southeast that are growing so rapidly? So way back when before all this was well known, Google did an initiative where we looked at our data centers which had been designed by the very best scientist according to us you know in our own arrogant way and we applied our own AI and it beat our own top people by 15%. That 15% of efficiency co uh went straight to the bottom line.
▶ 4:08:28It showed me that you can take any system and using AI do more what is called predictive analytics and you can predict loads and basically shed loads and handle it much more efficiently. That is where I our grid needs to be. Thank you. Uh Mr. Chairman, I have many more questions but I got 15 seconds. So with that, will you yield to me? Yes. I just say my purpose uh for Mr. Menendez for asking Dr. Smith to be here is not to come as a Republican witness and not tell us what we want to hear, but tell us what we need to hear.
▶ 4:08:56And I think we've all heard some things that probably don't fit within our aiology, but things we needed to hear and we can figure out and work through. So time's expired. I yield back and I will recognize I'm apologize to Miss Fletcher for missing her last time, but Miss Fletcher, you're recognized for five minutes. Thank you so much, Chairman Guthrie. I appreciate it. And um I appreciate all of our witnesses for being here today and for your testimony. Um I think this has been a really useful and important hearing.
▶ 4:09:20you've given us lots to think about and um we've heard from all of you right that the the United States is really on the brink of an AI revolution um that there are many things we need to be thinking about and just kind of the transformational change that this is going to bring including demand for energy and in normal times that should be great news for my home state of Texas uh where we already have a growing industry a cluster of data centers and we have the energy resources and the knowhow to um meet this
▶ 4:09:50sort of record high demand. Um, but President Trump's policies are eroding the certainty and predictability that the people who run businesses and make investments need to succeed at every turn. Um, and this is particularly true when it comes to building our infrastructure for our um, energy to meet tomorrow's demand. Um, so I want to focus a little bit on that, but Mr. Schmidt, I really appreciated your opening testimony today before the panel.
▶ 4:10:20Um, and I wrote down a few things that you were speaking about that I want to follow on. And you mentioned um, you referenced sort of the balance of power globally. And I think we can all acknowledge that we're in a very uncertain and shifting moment in our history. Uh, it's changing minutebym minute.
▶ 4:10:39And um but you said something I thought that was really important kind of I want to ask you about that in the context of um something you said in your written testimony which really struck me and I'm just quote from your testimony but you said the government can't win this technological race alone. We must reignite America's unique innovation power the potent collaboration between government private industry and academia. And I won't read the whole quote in the interest of time, but before I served on this committee, I served on the science, space, and technology committee.
▶ 4:11:06And I was struck at every single hearing by the witnesses. We always had a witness from academia, from the government, and from industry talking about how well and efficiently and effectively they collaborated. And so I assume that you would agree with me that the disruptions that we're seeing um are challenging in this moment. Um, I assume you would agree with me that regulatory certainty is an important factor for private industry and attracting capital and to projects. Yes.
▶ 4:11:35Um, and I assume you would agree that the supply chain disruptions and other kinds of things that we're experiencing are going to hurt productivity. Yes. Um, I also uh assume that you are aware based especially on your testimony about your involvement with the Mayo Clinic um that you're aware of the cuts to academic research that are happening um whether it's through the NIH and the the cost sharing uh for medical research or grant funding at various institutions.
▶ 4:12:02I keep hearing from my constituents in every industry that the increased uncertainty that we are experiencing as a result of this administration's policies, these are all new changes this year is really an impediment.
▶ 4:12:15And so I just want you to elaborate with the time we have left, there's about two minutes, on your vision for revitalizing the partnership that you described between industry and academia and the government and then share your thoughts on how we can and should do that in this environment and what kinds of changes we should make to make that possible. And I know we don't have everybody in the room today, but I've heard our colleagues on both sides of the aisle are listening and I think your insights here would be really important. Um, thank you.
▶ 4:12:44The Vanavar Bush post World War II constructed the the sort of structure that you're describing. The government is a regulator and a propos and a proponent and also does basic research funding. Universities do that research and then venture capital takes huge risks to do this. You see this in traditional democratic areas but also Republican areas.
▶ 4:13:07For example, fracking was an American invention following the same problem and it produced enormous benefits to America uh by virtue of economics and so forth. Everybody's aware of that. We're now essentially energy independent. So the role of innovation is core. I call this innovation power. I've written about this at some level. The future of America will be determined about the rate at which we can innovate. And we have unfortunately somebody who's trying to copy us and moves very quickly.
▶ 4:13:35Their innovation model is more centralized, but they're plenty smart. They got lots of resources and they're very focused and they do all the right things with respect to, of course, it's not a democracy. Getting the right smart people in the right place. They produce national champions as Alex mentioned and they push them and they push them hard for globalization. China is now in fact overbuilding manufacturing so that they can essentially essentially become the world's manufacturer again with huge impacts economically to everybody.
▶ 4:14:02You see the power of innovation right in front of you there in China. Why are we not going after that in AI? We should. We invented it. It's right in front of us. It is the core of everything we can do. New developments in physics and and biology and science and so forth. The current administration's cuts, the 15% indirect cost recovery, the NIH costs are not consistent with that vision. If they have a problem with specific programs, do it specifically, not generally. Thanks. Thank you so much. I have gone over my time.
▶ 4:14:31I really appreciate it and Guthrie, I yield back. Appreciate it very much. Gentle lady yields back. Mr. Evans, you're recognized for five minutes. Thank you, Mr. Chairman and ranking member and of course to the witnesses uh for for taking the time to testify today. Um uh Dr. Schmidt, my first question will be to you. Uh in your testimony, you talk a lot about China's investment in a lot of different forms of energy like wind, solar, and newer technologies like fusion. Um the United States has made similar uh investments in the past several years.
▶ 4:14:58Um but I think it's also important to highlight that not all energy is necessarily created equal. Um and so the first question to you is in your opinion uh which nation has brought more dispatchable base load energy generation online over the last 5 years between China and the US? Uh it's almost certainly China. And then in your opinion which nation has taken the most dispatchable base load energy offline in the last 5 years? Almost certainly the United States. Thank you.
▶ 4:15:27Um and and I agree with your answers there. Uh obviously I should include Germany for shutting down all of his nuclear plants, which is also a mistake. Thank you. Um but yeah, I agree with your answers there. Uh we know that China's thermal power generation has reached a record high just last year and that's driven by things like coal fired plants um which have also reached a a record high as a percentage of what it's generating in China.
▶ 4:15:52Um, and by comparison, uh, the United States is on track to retire 12.3 gawatts of dispatchable power this year. Um, and for me personally, that's concerning because over 10% of that, uh, about 1.3% of that, uh, is retirements of dispatchable base load power that's taking place in Colorado, uh, even though we're only 1.3% of total energy production in the United States.
▶ 4:16:19So with that focus on Colorado, uh the next question to you is we're taking over a gigawatt of power or scheduled to take a gigawatt of base load power offline in Colorado this year. 5 gawatts of dispatchable base load power offline by 2030 at the same time that my governor is saying he wants to make our state a quant excuse me a hub for quantum technology and and AI.
▶ 4:16:44So the question to you is if you wanted to be your state a make your state a leader in quantum computing AI um what would be the energy policy that you would want to see to support that? Um it probably makes sense to retire the coal plants and replace them by natural gas plants. Um it makes sense in Colorado because you have such great natural resources to work on enhanced geothermal.
▶ 4:17:10So there are things that you can do, but the core message I think from the entire panel here is we want more of everything, right? And that we want it sooner. And not only do we want it, we need it for American Thank you. And I appreciate your reference to the gas plants. Um because my district is truly an all of the above energy district. um 83% of the oil, 56% of the natural gas in Colorado, largest wind generating, you know, the wind turbine manufacturing facility.
▶ 4:17:38Um probably in the United States is headquartered in my district, geothermal, solar, we we truly are in all of the above. But specifically with gas plants, one of the things that I've heard there is that there's a major backlog in getting the the gas turbines. So, can you speak a little bit more to the timing of retiring coal generation if you don't have a a gas alternate immediately ready to go? I'm not enough of an expert to give you a precise answer.
▶ 4:18:02Um the reason the natural gas plants have be become more expensive is demand which is in is sort of what we want right we want more of everything and then the market will react. The problem is that these things take years to backlogs get you know that delay in natural gas plants will hurt AI competitiveness because it's the best source of power in certain situations. My my personal advice is start by since China is allegedly dumping solar panels, just buy them, right? Because they lower energy costs, right?
▶ 4:18:31Do whatever it takes to get more power into America as Mr. Mr. Turk says, more electrons. Thank you for that. Um Mr. Wang. Um, kind of, uh, pivoting off of that conversation, I'm just curious if you can speak to in my remaining 45 seconds just briefly, what happens if we lose this AI race with China? Uh, what what does the world look like if China becomes the leader in that space uh, and no longer the United States in part because we retired too much power?
▶ 4:19:03I I spoke to this and I think I think Dr. Schmidt made some uh relevant comments that you know AI is on the brink of becoming a very very powerful technology that is much more than just chat GPT. It's a reasoning engine. It has the ability to uh you know very soon conduct cyber attacks you know be be really an very important technology for national security.
▶ 4:19:24So to sum it up I guess in 10 seconds um you know in a world where the Chinese Communist Party wins uh they have clear intention to utilize AI as a mechanism to export their ideology globally as well as potentially uh you know enable them and other authoritarian countries to lead. Times expire on this. So thank you Mr. Evans. She yields back. We're trying to keep we have three more to go. Mr. Dr. Smith. So Mr. Lansman. Uh thank you Mr.
▶ 4:19:51chair and thank you uh to all of our panelists for your testimony today. This has been incredibly helpful and and you know the issue of of AI is one that you know we have to get right. There's there's no um there's no debate about that.
▶ 4:20:10Winning on AI and harnessing it for good requires as you all have said very I think impactfully clear instructions and guidance and meaningful investments. Mr. Turk, let me start with you. Congress has struggled to do this and I I I'm not picking a fight here. I'm not leading you in any direction. I'm I'm I'm genuinely curious.
▶ 4:20:39What do you think the barriers are in terms of us laying out that clear guidance and and and and making the necessary So the good news is uh Congress provided that certainty provided that window of investment.
▶ 4:20:58That's one of the brilliant parts of the legislation that you all passed on the tax credit side to have tax credits in place for 10 years that investors that developers that utilities that AI companies can rely upon and know will be there so that they can make investment decisions that will come to fruition over a period of years. So the good news is the biggest thing you have to do at this point is leave those tax incentives. Let that grant money do what the grant money uh was intended to do by Congress, but just execute on that.
▶ 4:21:27that that's on the investment and and so thank you for that on the investment and and piece of this but on the clear in instructions and guidance I mean what do you think is holding us back from providing that framework that everyone has been asking for well this is where uh the private sector will do what the private sector does best when it has that certainty it doesn't have the chaos from tariffs it doesn't have the chaos from repeal of provisions I also completely agree with all panelists.
▶ 4:21:57I don't think there's disagreement. Uh we need to build and we need to build quicker in this country, including transmission. Uh but a whole range of clean energy resources. Uh permitting takes too long in our country. It's complicated. We've made some progress on that, but we need to make more progress to make it durable. It needs to be bipartisan. And so I know there's conversations happening. We just need to get on with it. U Mr. Schmidt, thank you. I agree with that. Um Mr.
▶ 4:22:24I can you just talk a little bit about how important talent is? You you you discussed it earlier, but I how important talent is to this whole uh process and the impact of of of the chaos around the administration's immigration policies. So, Silicon Valley and the world I represent is powered by the smartest people or at least the self-proclaimed smartest people in the world.
▶ 4:22:52And we collectively need them because the algorithms and the approaches we take are in incomputable by normal people. I don't understand what most of these people are doing and I have a PhD in this area. Yeah, that's how complicated this stuff is. The new AI stuff is largely math um and it's a new set of math. In fact, there are people who are working on what are the limits of AI using again trying to find out where the really limits are.
▶ 4:23:19All of that knowledge is in the heads of people around the world who are highly specialized. They're not normal people. They're they're just geniuses in one way or the other, men and women. I want all of them here. It is insane to not let them in here. If you look at polymaths, I wrote a book on this called Genesis. We studied polymaths. A single polymath, the person who invents something, this is the Leonardo da Vinci type person can generate a trillion dollar industry.
▶ 4:23:47Carver me and so forth in the 1970s invented semiconductors now a multi- trillion dollar me industry we need those people in America imagine if each and every one of those people did not live in America they lived in another country in particular China furthermore we have lots of evidence for example that the quantum lead that China now has occurred because a specific quantum physicist was not allowed to stay in the country and he said okay I'll go back and work for China and the rest is history and quantum is a
▶ 4:24:17huge national security issue for America right now. Thank you for that. Uh also Dr. Schmidt just want to talk a little bit about prices. I only have a few seconds but um prices have gone up, electricity prices in Ohio and obviously this is going to cause even more um pressure on prices. Is it the tax credits and is is that the most important thing we can do to keep prices down or do you want Congress to do more? Um I want more supply.
▶ 4:24:47More supply should lead to better and tougher competition and more more dynamic network which allow would allow v vendor choice. The way we we the congress should not set prices. The congress should enable competition at every level in the value chain in every industry and in particular in electricity. Thanks. We have two more. Two more. If somebody else comes in after we're going to excuse you, Dr. Smith, we'll keep going. But Mr. Griffith is right. Thanks for yielding back. Mr. Griffith is recognized for Thank you very much. Let's continue talking about prices.
▶ 4:25:12So, it makes absolutely no sense to retire a coal plant, let's say, that was opened up in 2012 or 2013 that has a life expectancy of more than 50 years because we've decided we hate coal. Isn't that right, Mr. Schmidt? Dr. Schmidt. It has to do with how long you It's a more complicated answer. Okay. It has.
▶ 4:25:41So for the question, let me cut through some of the complications for the question that Mr. Lansman asked. He said our prices are going up. What do we need to do? You said we need more supply. I agree with that. But also, we can't leave stranded assets out there because this was opened up in my district in 2012 as the cleanest coal plant at the time uh in the world and a very clean plant and it's underutilized right now. And there's movements of foot to have it close up early.
▶ 4:26:11And that takes power away from us. And that affects prices because the consumer not only can't access the power because there's not enough supply, which you just said, but it also puts them in a situation where they're paying for the stranded asset of the existing coal plant and the new plant that might replace it. whe whatever fuel source it uses, whether it be nuclear, which I'm also in favor of, or whether it be natural gas, or whether it be uh wind or solar, that that's fairly straightforward, isn't it?
▶ 4:26:40Because if you leave the stranded asset, rateayers paying for both the old and the new. I grew up in coal country of Virginia, so I do understand it. Oh, what county? Uh Blackburg. Okay. Well, they did have coal mining there at one time, but yeah, I represent that area. That's my that's my district. Um the important thing about the important thing about coal is that over a over over the long run coal is going to get regulated out because it's such a it's coal is much dirtier than natural gas. You would always choose natural gas over coal if you made that decision today.
▶ 4:27:09Given that you have an under utilized coal plant, I would encourage you to look at the network interconnect. Why is it not fully used? Why are we not taking that resource that you described and fully using it right now? Because we're overregulating coal. I'll just answer that one for you. Now, I also have and this this gets to be interesting. I have an underutilized natural gas plant as well in the area. And so, we're trying to attract investment into that region that you grew up in. Blackburg is a wonderful town.
▶ 4:27:36I'm I also represent the coal fields where they still produce the coal and natural gas as well because of our coaled methane. And we've got a natural gas facility that used to be a coal facility was converted that's also underutilized. And we would love to see uh folks take a look because as you know having come from that region these are very industrious people and and whether or not they have that diploma I'm reminded of the scene in the Wizard of Oz.
▶ 4:28:04Uh there's a lot of smarts out there and I believe that both data centers and AI could benefit by being in the in the region. Uh but when you when you close down these facilities, now I understand you have a preference for natural gas and I I understand that.
▶ 4:28:18But when you close down these facilities, that creates a problem because wouldn't you agree we right now in last year the American Electric Reliability Corporation's long-term assessment estimated that 115 gawatts of dispatchable generation is planned to retire over the next 10 years in comparison to what they estimate to be an increased demand of 150 plus Doesn't that impede or or make it more difficult for us to have space to grow AI and power our AI as we need to?
▶ 4:28:48Again, I think all of us believe in more. Yep. With respect to the specifics, uh you have regulatory issues which you pointed out which I think should be loosened, but I also think the long-term for coal is to be replaced by natural gas and I think we should get organized around that and eventually natural gas will be replaced by fusion which will ultimately solve all of our problems 15 years from. I'm I'm looking forward to fusion getting here. I hope that you're right on your estimate.
▶ 4:29:12Uh I would say this as well because so many times people hear statements like that in my district and they automatically assume that that means coal production is going to end. And they don't realize that what you're talking about is coal coal production for the creation of electric generation. And my district has a rich seam, as you're probably aware, of metallurgical coal, which for those who don't know means that that we mine that coal to make coke and steel out of it so that we can produce the steel that's needed for this country.
▶ 4:29:40I think somebody mentioned it earlier today that we need the steel so we can make sure we build the equipment and so forth to do the AI with the buildings, etc. And you're not going to make that that really good steel without burning some of my metallurgical coal. I yield back, Mr. Chairman. Gentleman yields back and the chair recognizes Miss Mlullen. Uh, thank you, Chair. Thank you, Chairman Guthrie and member Palone for planning this hearing. This is probably my favorite hearing of my entire almost a little over two-year congressional career.
▶ 4:30:10It is definitely the most important. And Dr. Schmidt, I'm glad you stayed because in your opening statement, you said that the sheer speed of AI development is outpacing our societal and government ability to adapt. And I wholeheartedly agree with you. In fact, seven years ago in 2018, I attended a conference at which a speaker was talking about the rise of AI and mega trends and all of these things and he basically said the same thing.
▶ 4:30:37And I came to the conclusion seven years ago that none of our systems in the United States at all, government, education, none of them are prepared for what's coming. But at the same time, as Mr. Wang testified at that point 7 years ago, China already had an AI master plan, advanced capabilities, and President Gping declared China's plan to dominate AI by 2030.
▶ 4:31:04Yet, this committee held its first hearing on AI in 2023. The race for AI dominance is reminiscent of the space race. Uh, but instead of the Soviet Union, now it's China, but the stakes are even higher. And we won the race to land a man on the moon. And that was critically important to our economy and our national security and innovation and scientific advancement. And to win the race for AI is just as important.
▶ 4:31:35But as Mr. Wang testified, while the US leads on computing and we're tied with China on algorithmic development, China leads on data, which is the raw material that enables AI to learn, adapt, and improve over time. And as Mr. uh Wang said is AI's oil, gas, wind, solar, all wrapped in one. So if we lose the race to lead data, we lose the race for AI dominance.
▶ 4:32:02Now, the Trump administration's actions since January 20th have directly undermined our ability to win the race for AI dominance. The haphazard firing of federal workers, freezing or cutting federal funds for government agencies and universities critical to supporting competing AI hinders our ability to implement the imp the recommendations of Mr. Wang's testimony and his four pillars to win.
▶ 4:32:27This war on renewables that the president is engaged in attempts to repeal the inflation reduction act tax credits undermine the ability to meet our energy demands for data centers. And Trump's reckless tariff policy is increasing costs, exacerbating supply and demand issues already occurring, and raising the cost to build new data centers and semiconductor manufacturing plants that are critical for our AI success.
▶ 4:32:53Because while semiconductors have been exempted from the tariffs, the equipment and machinery used to build and run the data centers have not. This is not theoretical. Just this week, Microsoft announced that it's backing off plans to build three data centers in Ohio.
▶ 4:33:10So, given this committee's clear desire to position the US to win the competition with China for AI dominance, I am perplexed by my colleagues on the other side of the aisle silence over the Trump administration's actions that hinder our ability to do so. And the blank check, it looks like we are about to give the president to take those actions. So, Mr.
▶ 4:33:34Wang, you offer two options for AI dominance in the future and recommend working with our allies to promote an American model of AI technology. But this trade war is actively undermining our ability to work with our allies to do so. Can you elaborate quickly on the steps that we should actually take to work with our allies to promote a US model of AI usage and governance? Yes.
▶ 4:34:04So, uh the first thing is we need to ensure that NIST uh the National Institute of um standards uh is properly resourced and we're able to make progress on AI measurement science and ultimately the development of these AI standards.
▶ 4:34:19Um then we need to codify this into a set of standards that uh that we ultimately agree with in terms of how we should measure AI performance, how we should uh what what are the characteristics of safe safe and performant AI systems in the future and then we should utilize the global network of AI safety institutes which is uh which already exists. Many many countries have stood them up. Um you know France, the UK, uh Japan, India, Korea.
▶ 4:34:47I've met the the heads of many of these um AI safety institutes. They're all looking uh towards the United States because um you know they understand that we are the leader in the technology and we need to uh give them our standards and export it globally. Thank you and I yield back. Thank you. The gentle lady time I had committed to Dr. Smith that he got we have Dr. Shrier did come in but I committed to Dr. Smith to leave you Dr. Dr.
▶ 4:35:13I recognized for five minute and if anybody else shows up you're well you'll let you walk up and go. Thank you for your because it's been valuable. I I'll shut up and let her go. I'm so glad you're staying. Um let's see. Uh thank you uh thank you Mr. Chairman. I I and thank you to all our witnesses. This is a fantastic hearing.
▶ 4:35:34Uh I I'm from the Pacific Northwest and chip manufacturing and data center expansion are the big uh energy demand drivers to the region. So I'm thrilled to have this discussion. We're at this inflection point. We all know uh that we are really headed straight to an an energy crisis if we don't act quickly on this.
▶ 4:35:56It impacts AI and data centers as we've heard a lot about, but also we've been talking nationally a lot about manufacturing and we need affordable energy for that. Um, one of the best ways to ma to maximize access to the power we already have in the US is strategically building out transmission. And last year, Senators Mansion and Baraso introduced the bipartisan energy permitting reform act. Um, and I'll be really clear, it's not the bill I would have written.
▶ 4:36:24Um, I was not a fan of all the provisions, but we need to move forward, and that's the whole idea that we need compromise in order to move the ball forward. Um, Mr. Bota, in your testimony, I see this prime example that you've talked about a couple places, a couple times where this bill for speeding permitting would make a difference. It was the the Boardman to Hemingway transmission project that connects uh Oregon to Idaho.
▶ 4:36:50And in the Pacific Northwest, uh, our peak energy demands in the winter when we turn on the heat. Uh, and yet we have our peak hydropower generation in the warmer months when the snow melts. The opposite is true in the mountain region where we see the opposite. So irrigation and air conditioning drive that demand in the summer in the summer and then wind energy is more abundant in the spring and winter.
▶ 4:37:17So connecting those two regions uh would allow us to correct this mismatch and meet the demand. The project as you said is about to hit its 21st birthday and it has been stalled for almost 21 years. Um if we continue to require transmission projects to jump through all of these hoops and red tape, how is that going to hamper our ability to onshore tech, keep on tech, and uh expand manufacturing here at home?
▶ 4:37:46Um absolutely u you know EPRA is something we're absolutely uh supportive of because what it's going to do is exactly what you mentioned you know we've talked about um investing in the grid we've talked about modernizing grid creating more flexibility so that you can balance supply and demand um and you know the the big data centers certainly the large semiconductor manufacturing which we are under construction right now in in in Boise of uh what will be the largest the only large scale memory manufacturing facility in the in the country the first leading edge one in in 25
▶ 4:38:17more than 25 years um needs that transmission to be able to ensure that we can have that stable power for the consistent and long-term load growth that we have. And Dr. Schmidt um basically same question. If we don't have good transmission and the ability to move uh energy across the country, how does that impact our ability to remain dominant and win the AI race? When I think about your state, I think about all of the incredible natural resources you have.
▶ 4:38:44um whether it's the west or east part of your state um that power does not have the path out of your state that's strong enough. It needs to get fixed. Thank you. Um I I just want to emphasize for my Republican colleagues that if they introduce a bill like that one, they will have uh Democratic partners because we all understand, especially after this hearing, that we need to get just add one more just one more thing to add to this. It's not it's not just about the success of those projects.
▶ 4:39:13I know we're talking a lot about AI but it's about jobs that are being that that all of this investment in manufacturing are going to be creating high-paying jobs higher paying jobs in today and domestic supply of semiconductors while while critical and important for AI is also critical for many many other industries that we haven't been able to talk about automotive industry for example 50% of the cars on the road have a chip made in Micron's facility that's right and we need to make so many many industries that need these projects to be successful.
▶ 4:39:43I'm going to turn one more question quickly to do to to Dr. Schmidt. Um, AI, as I read in Dr. Wang's testimony, Mr. Wang's testimony, brings potential benefit, potential risk. We've seen the abuse of AI in China for public surveillance and crackdowns.
▶ 4:40:00Now, unfortunately, I'm having to think about that in our country, too, with uh what we're seeing now as suppression of disscent and retribution uh efforts to uh to crack down on free speech and sty scientific research target nonviolent university protesters who I may not agree with, but we all have the first amendment rights. And um we've also seen insurance companies uh with Medicare Advantage use AI to deny or delay coverage.
▶ 4:40:28So, as you think about I we only have we actually don't have any time. If you could write me an answer to what you would suggest for guard rails for AI as we move forward. We want to be able to keep up and do this wisely. Um I will I will do so. Thank you very much. Thank you. All right. Seeing no further uh folks here to ask questions, ask unanimous consent to insert in the record the documents included on the staff hearing documents list. Without objection, that will be the order.
▶ 4:40:58I remind members they have 10 business days to submit questions for the record and I ask the witnesses to respond to the questions promptly. Without objection, committee is adjourned.