Powering Demand: Nuclear Solutions for AI Infrastructure

1[House Hearing, 119 Congress]2[From the U.S. Government Publishing Office]34                            POWERING DEMAND:5                NUCLEAR SOLUTIONS FOR AI INFRASTRUCTURE6=======================================================================7                                HEARING89                               BEFORE THE1011                         SUBCOMMITTEE ON ENERGY1213                                 OF THE1415                      COMMITTEE ON SCIENCE, SPACE,16                             AND TECHNOLOGY1718                                 OF THE1920                        HOUSE OF REPRESENTATIVES2122                    ONE HUNDRED NINETEENTH CONGRESS2324                             FIRST SESSION2526                               __________2728                             JUNE 12, 20252930                               __________3132                           Serial No. 119-153334                               __________3536 Printed for the use of the Committee on Science, Space, and Technology3738 [GRAPHIC NOT AVAILABLE IN TIFF FORMAT]3940         Available via the World Wide Web: http://science.house.gov4142                                __________4344                U.S. GOVERNMENT PUBLISHING OFFICE4560-594PDF              WASHINGTON : 202646=======================================================================4748              COMMITTEE ON SCIENCE, SPACE, AND TECHNOLOGY4950                   HON. BRIAN BABIN, Texas, Chairman51RANDY WEBER, Texas                   ZOE LOFGREN, California, Ranking52JIM BAIRD, Indiana                       Member53DANIEL WEBSTER, Florida              SUZANNE BONAMICI, Oregon54JAY OBERNOLTE, California            HALEY STEVENS, Michigan55CHUCK FLEISCHMANN, Tennessee         DEBORAH ROSS, North Carolina56DARRELL ISSA, California             ANDREA SALINAS, Oregon57CLAUDIA TENNEY, New York             VALERIE FOUSHEE, North Carolina58SCOTT FRANKLIN, Florida              EMILIA SYKES, Ohio59MAX MILLER, Ohio                     MAXWELL FROST, Florida60RICH McCORMICK, Georgia              GABE AMO, Rhode Island61MIKE COLLINS, Georgia                SUHAS SUBRAMANYAM, Virginia62VINCE FONG, California               LUZ RIVAS, California63DAVID ROUZER, North Carolina         SARAH McBRIDE, Delaware64KEITH SELF, Texas                    LAURA GILLEN, New York65PAT HARRIGAN, North Carolina         GEORGE WHITESIDES, California,66SHERI BIGGS, South Carolina               Vice-Ranking Member67JEFF HURD, Colorado                  LAURA FRIEDMAN, California68MIKE HARIDOPOLOS, Florida            APRIL McCLAIN DELANEY, Maryland69MIKE KENNEDY, Utah                   JOSH RILEY, New York70NICK BEGICH, Alaska                  BILL FOSTER, Illinois71VACANT72                                 ------7374                         Subcommittee on Energy7576                   HON. RANDY WEBER, Texas, Chairman77JIM BAIRD, Indiana                   DEBORAH ROSS, North Carolina,78CHUCK FLEISCHMANN, Tennessee             Ranking Member79CLAUDIA TENNEY, New York             ANDREA SALINAS, Oregon80PAT HARRIGAN, North Carolina         LAURA FRIEDMAN, California81SHERI BIGGS, South Carolina          JOSH RILEY, New York82JEFF HURD, Colorado                  VALERIE FOUSHEE, North Carolina83NICK BEGICH, Alaska8485                        C  O  N  T  E  N  T  S8687                             June 12, 20258889                                                                   Page9091Hearing Charter..................................................     29293                           Opening Statements9495Statement by Representative Randy Weber, Chairman, Subcommittee96  on Energy, Committee on Science, Space, and Technology, U.S.97  House of Representatives.......................................     998    Written Statement............................................    1099100Statement by Representative Deborah Ross, Ranking Member,101  Subcommittee on Energy, Committee on Science, Space, and102  Technology, U.S. House of Representatives......................    11103    Written Statement............................................    13104105Statement by Representative Brian Babin, Chairman, Committee on106  Science, Space, and Technology, U.S. House of Representatives..    14107    Written Statement............................................    16108109Statement by Representative Zoe Lofgren, Ranking Member,110  Committee on Science, Space, and Technology, U.S. House of111  Representatives................................................    17112    Written Statement............................................    17113114                               Witnesses:115116Mr. Pat Schweiger, Chief Technology Officer, Oklo117    Oral Statement...............................................    18118    Written Statement............................................    21119120Ms. Kathleen L. Barron, Executive Vice President and Chief121  Strategy and Growth Officer, Constellation Energy122    Oral Statement...............................................    26123    Written Statement............................................    29124125Dr. Jeremy Renshaw, Executive Director AI & Quantum, Electric126  Power Research Institute (EPRI)127    Oral Statement...............................................    35128    Written Statement............................................    37129130Discussion.......................................................    45131132              Appendix: Answers to Post-Hearing Questions133134Mr. Pat Schweiger, Chief Technology Officer, Oklo................    70135136Ms. Kathleen L. Barron, Executive Vice President and Chief137  Strategy and Growth Officer, Constellation Energy..............    74138139Dr. Jeremy Renshaw, Executive Director AI & Quantum, Electric140  Power Research Institute (EPRI)................................    76141142                            POWERING DEMAND:143                           NUCLEAR SOLUTIONS144                         FOR AI INFRASTRUCTURE145146                              ----------147148                        THURSDAY, JUNE 12, 2025149150                  House of Representatives,151                            Subcommittee on Energy,152               Committee on Science, Space, and Technology,153                                                   Washington, D.C.154155    The Subcommittee met, pursuant to notice, at 10:02 a.m. in156room 2318, Rayburn House Office Building, Hon. Randy Weber157[Chairman of the Subcommittee] presiding.158[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]159160    Chairman Weber. The Committee will come to order; the161Subcommittee on Energy is convening.162    Without objection, the chair is authorized to declare163recesses of the Subcommittee at any time.164    Welcome to today's hearing entitled, ``Powering Demand:165Nuclear Solutions for AI (Artificial Intelligence)166Infrastructure.'' I recognize myself for 5 minutes for an167opening statement.168    So listen, good morning everybody. We're glad you all are169here. That includes the audience, by the way. Welcome to the170hearing titled, ``Powering Demand: Nuclear Solutions for AI171Infrastructure.''172    Folks, with artificial intelligence's rapidly growing173demand on our power grid, this hearing is going to examine the174U.S. energy landscape and our capacity to meet that need. In175light of recent announcements around the country, we are going176to focus on nuclear energy's role as a baseload power source.177We will also review the Department of Energy (DOE's) research,178development and demonstration programs supporting the next179generation of nuclear reactors. Much like the 1940s--most of180y'all weren't here then and, just for the record, neither was181I--much like the 1940s, America stood at the dawn of the atomic182era, we now stand on the threshold of a new age, driven by183artificial intelligence.184    With promises of increased efficiency and productivity, AI185has the potential to revolutionize every single aspect of our186economy and our way of life. Unsurprisingly, this potential has187spurred a major influx of private capital aiming to turn these188very promises into our reality.189    Additionally, this technology is poised to dramatically190transform our electric grid and our energy sector due to the191construction of those very same AI data centers. According to a192recent report from a leading consulting company, data centers--193do you all say data or data?194    Data. There is one vote for data right there in the back.195All right.196    According to one consulting company, data centers are197projected to consume 5.2 percent of U.S. electricity this year,198with that share expected to increase to 11.7 percent by the199year 2030. Their energy use would rise from 25 gigawatts to 80200gigawatts. Let's put that in perspective: 1 gigawatt of energy201equates to roughly 294 utility-scale wind turbines, 1.8 million202solar panels, 103 offshore wind turbines, or 1 large light203water nuclear reactor, which is why we're here. I see your204excitement down there.205    Due to these immense energy demands, major technology206companies and hyperscalers who traditionally sit on the energy207sidelines are now climbing into the driver's seat to secure208their long-term power supply. Nuclear has emerged as the ideal209energy source, given its clean baseload power and unmatched210reliability. Nuclear's capacity factor of 92.5 percent--let211that sink in--is the highest of any, any energy source. This212level of reliability is essential for data centers, which can213afford no more than 5.25 minutes of downtime annually. No more214than 5.25 minutes downtime annually.215    As a result, tech companies are making substantial216investments and forging strategic power purchase agreements217with nuclear designers. For instance, Amazon has invested over218$300 million in X-energy. Google has signed a 500 megawatt219power purchase agreement with Kairos Power, and Switch secured220up to 12 gigawatts from Oklo Energy, another witness here221today, to partner with Microsoft to recommission a nuclear222reactor at the Three Mile Island site.223    In addition to investments and partnerships, tech firms are224also redesigning interconnection agreements. Traditionally,225data centers relied on the front-of-the-meter interconnection226agreements, in which a utility generates, transmits, and then227distributes energy to the end user. Now tech companies are228pursuing behind-the-meter-agreements where energy assets229directly power the data center. This approach reduces230transmission costs, streamlines the approval process, and231provides for greater operational control to the end user.232    Last year Talen Energy sold Amazon a data center and agreed233to a behind-the-meter power agreement, supplying up to 480234megawatts from its Susquehanna Nuclear Power Plant. Although235FERC (Federal Energy Regulatory Commission) blocked this236specific agreement, I believe back-of-the-meter interconnection237agreements will be commonplace when we're constructing new238nuclear power plants.239    As the nuclear and tech sectors deepen their collaboration,240the Department of Energy, DOE, Office of Nuclear Energy plays a241crucial role in AI's future success. The office manages the242Advanced Reactor Demonstration Program (ARDP) which helps243commercialize new nuclear technologies, including X-energy's244XE-100 reactor. Just recently Dow and X-energy submitted their245construction permit application for a proposed project in246Seadrift, Texas, which is right south of our district--my247district.248    DOE also manages the Nuclear Fuel Security Program and the249Advanced Nuclear Fuel Availability Program, both created under250the Energy Act of 2020. These initiatives are critical to251deploying next-generation reactors that rely on high-assay,252low-enriched uranium, or HALEU, fuel. DOE recently announced it253will provide help to X-energy, Kairos Power, Radiant254Industries, Westinghouse, and TerraPower.255    In tandem, the Trump Administration has accelerated the256deployment of AI through recent executive orders (EOs). In257April DOE issued a request for information, RFI, soliciting258feedback on whether industry would be interested in using259Federal land at DOE sites to host AI data centers. Given the260strong support for DOE and its nuclear security missions, these261sites are perfect, perfect for hosting data centers powered by262nuclear reactors. From the atomic bomb to artificial263intelligence, DOE's labs have consistently stood at the264precipice of technological advancement. They continue to meet265generational challenges and drive innovations that strengthen266America as we enter this new area.267    [The prepared statement of Chairman Weber follows:]268269    Good morning. Welcome to today's Energy Subcommittee270hearing titled, ``Powering Demand: Nuclear Solutions for AI271Infrastructure.'' With artificial intelligence's rapidly272growing demand on our power grid, this hearing will examine the273U.S. energy landscape and its capacity to meet that need. In274light of recent announcements around the country, we will focus275on nuclear energy's role as a baseload power source. We will276also review the Department of Energy's research, development,277and demonstration programs supporting the next generation of278nuclear reactors.279    Much like the 1940s, when America stood at the dawn of the280atomic era, we now stand on the threshold of a new age driven281by artificial intelligence. With promises of increased282efficiency and productivity, AI has the potential to283revolutionize every aspect of our economy and way of life.284Unsurprisingly, this potential has spurred a major influx of285private capital aiming to turn these promises into reality.286    Additionally, this technology is poised to dramatically287transform our electric grid and energy sector due to the288construction of new AI data centers. According to a recent289report from a leading consulting company, data centers are290projected to consume 5.2% of U.S. electricity this year, with291that share expected to increase to 11.7% by 2030. Their energy292use would rise from 25 GW to 80 GW. To put that in perspective,2931 GW of energy equates to roughly 294 utility-scale wind294turbines, 1.8 million solar panels, 103 offshore wind turbines,295or one large light-water nuclear reactor.296    Due to these immense energy demands, major technology297companies and hyperscalers, who traditionally sat on the energy298sidelines, are now climbing into the driver's seat to secure299their long-term power supply. Nuclear has emerged as the ideal300energy source given its clean baseload power and unmatched301reliability. Nuclear's capacity factor of 92.5% is the highest302of any energy source. That level of reliability is essential303for data centers, which can afford no more than 5.25 minutes of304downtime annually.305    As a result, tech companies are making substantial306investments and forging strategic power purchase agreements307with nuclear designers. For instance, Amazon invested over $300308million in X-Energy, Google signed a 500 MW power purchase309agreement with Kairos Power, and Switch secured up to 12 GW310from Oklo, which is one of our witnesses here today. This led311Constellation Energy, another witness here today, to partner312with Microsoft to recommission a nuclear reactor at Three Mile313Island.314    In addition to investments and partnerships, tech firms are315also redesigning interconnection agreements. Traditionally,316data centers relied on front-of-the-meter interconnection317agreements, in which a utility generates, transmits, and318distributes energy to the end user. Now, tech companies are319pursuing behind-the-meter agreements, where energy assets320directly power the data center. This approach reduces321transmission costs, streamlines the approval processes, and322provides greater operational control to the end user.323    Last year, Talen Energy sold Amazon a data center and324agreed to a behind-the-meter power agreement supplying up to325480 MW from its Susquehanna nuclear power plant. Although FERC326blocked this specific agreement, I believe back-of-the-meter327interconnection agreements will be commonplace when328constructing new nuclear power plants.329    As the nuclear and tech sectors deepen their collaboration,330the Department of Energy's (DOE) Office of Nuclear Energy plays331a crucial role in AI's future success. The office manages the332Advanced Reactor Demonstration Program, which helps333commercialize new nuclear technologies, including X-Energy's334Xe-100 reactor. Just recently, Dow and X-Energy submitted their335construction permit application for a proposed project in Sea336Drift, Texas, right next to my district.337    DOE also manages the Nuclear Fuel Security Program and the338Advanced Nuclear Fuel Availability Program, both created under339the Energy Act of 2020. These initiatives are critical to340deploying next-generation reactors that rely on high-assay low-341enriched uranium (HALEU) fuel. DOE recently announced it will342provide HALEU to X-Energy, Kairos Power, Radiant Industries,343Westinghouse, and TerraPower. In tandem, the Trump344Administration has accelerated the deployment of AI through345recent executive orders.346    In April, DOE issued a Request for Information (RFI)347soliciting feedback on whether industry would be interested in348using federal land at DOE sites to host AI data centers. Given349the strong support for DOE and its nuclear security missions,350these sites are perfect for hosting data centers powered by351nuclear reactors.352    From the atomic bomb to artificial intelligence, DOE's labs353have consistently stood at the precipice of technological354advancement. They continue to meet generational challenges and355drive innovations that strengthen America as we enter this new356era.357    I look forward to our discussion here today and yield back358the balance of my time.359360    Chairman Weber. I look forward to our discussion today, and361I'm going to yield the balance of my time to the gentlelady to362my right, the Ranking Member.363    Ms. Ross. Well, thank you very much, Chairman Weber, for364convening this hearing today to discuss the potential for365nuclear energy solutions to power our Nation's growing366artificial intelligence infrastructure.367    I also want to thank our distinguished witnesses for being368here to share your testimony and insights on this topic.369    We've heard repeatedly from a wide range of sources about370how the advancement of AI could fuel economic growth,371geopolitical advantages, and the acceleration of science and372technology. I'm proud to represent North Carolina's 2d373District, where AI is driving a rapidly growing job market,374education and workforce development, and research and375innovation that could play a significant role in ensuring our376Nation's competitive edge.377    Just over a month ago an organization in the Research378Triangle Park added a cutting-edge AI tool for clinical trial379optimization, leading to fewer required patients per trial,380requiring--reducing timelines and costs, and ultimately381increasing success rates. However, these advancements depend on382our ability to power them, and that's why we're here today.383    A recent assessment from SemiAnalysis found that the energy384needed to meet this infrastructure demand is projected to385require 80 gigawatts of additional energy by 2030. This need is386not foreign to me. Data center construction rose 15 times in387North Carolina last year. AI data centers are particularly388unique in their energy requirements, with some calling for 24/7389supply and massive--at a massive scale, and that scale is only390projected to grow as hyperscalers invest billions of dollars391toward these centers.392    Much like the promise of AI, these data center investments393can accelerate our economy and provide workforce benefits for394years to come, but not without a cost. There are many impacts395to consider in trying to meet this energy demand.396    First, the environmental cost of rapid data center397expansion are--that's far from negligible. We've been hearing398about this in the western part of North Carolina. A report from399DOE's Lawrence Berkeley National Laboratory released last year400said that the total greenhouse gas emissions for U.S. data401centers in 2023 was estimated to be 61 billion kilograms of402CO2 equivalent. As a former clean energy lawyer, I403know firsthand that meeting this energy demand while working to404build a sustainable future for North Carolina and the Nation is405not an easy task. We must intentionally build a more reliable406and sustainable energy supply, and we should be using any407environmentally benign resources, including nuclear, available408to do so at the lowest cost possible for the American people.409    We must also consider using existing under-utilized sources410of power, and must be intentional with our siting for data411centers required for this type of development. Siting data412centers near existing power generation, including in areas413where manufacturing has moved away, should remain top of mind.414And this has happened in North Carolina.415    Now, while not directly connected to a specific data center416by a dedicated power line, the Yadkin River Hydroelectric417Project in western North Carolina plays a role in the overall418power grid that served the data center that is in this Hickory419corridor all around that part of North Carolina. This area has420seen a significant investment in data centers, and the region's421hydroelectric power generation contributes to the overall422energy supply. We can and should use renewable and existing423energy sources to help power our data centers across the United424States.425    That also brings another point up: the cost passed down to426the American people of constructing new energy sources, the427ratepayers. They should not be first in line to absorb high428risk and financially burdened with typical--with the typical429rates associated with deploying a first-of-kind energy source.430    Today we're here to discuss nuclear power, an energy source431with immense promise, but with a very high risk of cost and432schedule overruns thus far. So we must keep top of mind who433will absorb this risk. I don't have to look far to see the434negative impacts that the risk being absorbed by the ratepayers435can--where it can arise. The abandoned VC Summer project in436South Carolina experienced years of delays and cost overruns,437eventually resulting in $9 billion of sunk costs. Now, we know438some other people are looking at taking that over, but that439does not get rid of what the ratepayers had to absorb. I'm glad440I'm seeing some nodding heads over there.441    To this day, $5.7 billion is still being paid for by442customers who do not reap the benefits of a successful nuclear443project. This contrasts sharply with the very positive444experience in Georgia with their Vogtle project, which was445enabled by strategic support from DOE's Loan Program Office, or446LPO. This is exhibit A for the reason we absolutely must447maintain expanded resources and capabilities to provide to the448LPO through the Inflation Reduction Act, and those are at risk449right now.450    Nuclear projects are inherently challenging, but I believe451we can learn from past mistakes and proceed responsibly452considering the use of brownfields, Federal land, strategic453capital, and, above all, the protection of the ratepayers while454building a more reliable and sustainable energy supply. We have455an opportunity to build long-lasting infrastructure that456unlocks a more reliable and sustainable future, which I intend457to continue working toward with the Chairman. I look forward to458hearing from our witnesses today on how nuclear energy can help459us get there, in tandem with a far broader portfolio of clean460energy resources that should also be considered.461    [The prepared statement of Ms. Ross follows:]462463    Good morning and thank you, Chairman Weber, for convening464this hearing today to discuss the potential for nuclear energy465solutions to power our nation's growing artificial intelligence466infrastructure. I also want to thank our distinguished467witnesses for being here to share your testimony and insights468on this topic.469    We have heard repeatedly from a wide range of sources about470how the advancement of AI could fuel economic growth,471geopolitical advantages, and the acceleration of science and472technology. I am proud to represent North Carolina's 2nd473District, where AI is driving a rapidly growing job market,474education and workforce development, and research and475innovation that could play asignificant role in ensuring our476nation's competitive edge.477    Just over a month ago, an organization in Research Triangle478Park added a cutting-edge AI tool for clinical trial479optimization, leading to fewer required patients per trial,480reduced timelines and costs, and ultimately increased success481rates. However, these advancements depend on our ability to482power them.483    A recent assessment from SemiAnalysis found that the energy484needed to meet this infrastructure demand is projected to485require 80 gigawatts of additional energy by 2030. This need is486not foreign to me: data center construction rose 15x in North487Carolina last year. AI data centers are particularly unique in488their energy requirements, with some calling for 24/7 supply at489massive scale--and that scale is only projected to grow as490hyperscalers invest billions of dollars towards these centers.491    Much like the promise of AI, these data center investments492can accelerate our economy and provide workforce benefits for493years to come--but not without a cost. There are many impacts494to consider in trying to meet this energy demand.495     First, the environmental costs of rapid data center496expansion are far from negligible. In a report from DOE's497Lawrence Berkeley National Laboratory released last year, the498total greenhouse gas emissions for U.S. data centers in 2023499was estimated to be 61 billion kilograms of CO2500equivalent.501    As a former clean energy lawyer, I know firsthand that502meeting this energy demand while working to build a sustainable503future for North Carolina, and the nation, is not an easy task.504We must intentionally build a more reliable and sustainable505energy supply, and we should be using any environmentally506benign resources available to do so at the lowest cost possible507for the American people. We must also consider using existing,508underutilized sources of power and must be intentional with our509siting for the data centers required for this type of510development.511    Siting data centers near existing power generation-512including in areas where manufacturing has moved away--should513remain top of mind. While not directly connected to a specific514data center by a dedicated power line, the Yadkin River515Hydroelectric Project in western North Carolina plays a role in516the overall power grid that served the data center corridor517around Hickory.518    This area has seen significant investment in data centers,519and the region's hydroelectric power generation contributes to520the overall energy supply. We can and should use renewable and521existing energy sources to help power our data centers across522the United States.523    That brings us to another point we must keep top of mind524today: the cost passed on to the American people--the525ratepayers--who should not be the first in line in absorbing526the high risk and financial burden typically associated with527deploying a first of a kind energy source. Today we are here to528discuss nuclear power--an energy source with immense promise to529be sure, but with a very high risk of cost and schedule530overruns thus far.531    So, we must keep top of mind: who will absorb that risk?532    I don't have to look far to see the negative impacts of533that risk being absorbed by taxpayers. The abandoned VC Summer534project in South Carolina experienced years of delays and cost535overrun, eventually resulting in $9 billion of sunk cost. To536this day, $5.7 billion is still being paid for by customers who537do not reap the benefits of a successful nuclear project. This538contrasts with Georgia's Vogtle project, which was enabled by539strategic support from DOE's Loan Programs Office, or LPO.540    This is Exhibit A for the reason we absolutely must541maintain the expanded resources and capabilities provided to542LPO through the Inflation Reduction Act. Nuclear projects are543inherently challenging, but I believe we can learn from our544mistakes and proceed responsibly--considering the use of545brownfields, federal land and strategic capital, and above all546the protection of taxpayers while building a more reliable and547sustainable energy supply.548    We have an opportunity to build long-lasting infrastructure549that unlocks a more reliable and sustainable future, which I550intend to continue working towards.551    I look forward to hearing from our witnesses today on how552nuclear energy can help us get there, in tandem with the far553broader portfolio of clean energy resources that should also be554considered.555    Thank you, and I yield back.556557    Ms. Ross. Thank you, and I yield back.558    Chairman Weber. Thank you, Ranking Member Ross. I now559recognize the Chairman of the Full Committee, Chairman Babin.560    Chairman Babin. Yes, sir. Thank you very much, Mr.561Chairman, and thank you to our witnesses today for being here.562I want to thank our Energy Subcommittee Chairman, my good563friend, Mr. Weber, for holding this timely hearing on powering564data centers with advanced nuclear technologies.565    Artificial intelligence is advancing at a breakneck pace.566Every day we see brand new, groundbreaking developments and567headlines that highlight AI's expanding role in our economy,568our national security, and even our daily lives. Fortunately,569the United States currently leads the world in the AI race, but570that lead is not guaranteed. Both friends and foes are moving571swiftly to close this gap.572    At the core of AI innovation are data centers, the engines573that drive this technological revolution. These facilities574require enormous and uninterrupted power. And when it comes to575reliability, only one option consistently meets those high576standards, and that is nuclear.577    Industry standards for top tier data centers, especially578those supporting AI workloads, require what's known as the579``five-9s'' uptime, or 99.999 percent operational reliability.580This means that over the course of a year the data center is581expected to be operational 99.999 percent of the time. That582equates to just 5.2 minutes of allowable downtime per year. Few583energy sources can meet that bar. In fact, nuclear power, with584its 92.5 percent capacity factor, stands alone in its ability585to provide the clean, constant baseload power that these586systems demand. Plus, nuclear can provide much more energy per587square foot than its competing energy sources, allowing nuclear588to be sited nearer to its customers and be less impactful on589overall land usage.590    As this Administration has stated, nuclear energy will play591a central role in our energy future. Thanks to executive orders592from President Trump and recent actions from the Department of593Energy, we are beginning to cut through the red tape that has594long stalled nuclear progress, moving quickly but safely to595scale nuclear power nationwide. To get to this point, the596Federal Government has spent over a decade significantly597investing in advanced nuclear through DOE's Office of Nuclear598Energy. These early stage investments were essential to getting599first-of-a-kind technologies off of the ground.600    However, in today's budget-constrained environment, that601support must be seen as a launch pad, not a permanent lifeline.602It's time for the private sector to stand up--I should say step603up. Fortunately, the market is ready. The nuclear renaissance604we anticipated in the early 2000s is now within reach. As the605demand for power increases, new nuclear companies have a606critical partner in large technology firms that are willing to607invest in and help deploy the fleet of nuclear reactors that608has long been promised. However, these tech companies are not609only willing to pay a premium to bring these new, first-of-a-610kind power sources online, but they have also shown an appetite611for current technology.612    Earlier this year Microsoft and Constellation Energy, one613of our witnesses today, announced plans to reactivate the Crane614Clean Energy Center, and formerly Three Mile Island, marking a615historic moment. Additionally, just last week Meta and616Constellation announced a 20-year agreement to keep the plant617operational. This isn't just good energy policy. It's good618economic policy, creating well-paying jobs and generating619millions in tax revenue that would otherwise be lost.620    This momentum is spreading across the country, including my621home State of Texas. Lawmakers are advancing pro-nuclear622policies. Universities are also seeing the benefits of623partnering with advanced nuclear. Texas A&M, through its Rellis624Campus, announced the creation of the Energy Proving Ground625Project. This project will involve Texas A&M assisting four626chosen SMR (small modular reactor) companies with permitting,627and providing a location to demonstrate their reactors.628    Now is the time for the nuclear industry to take its next629step toward unlocking its full potential. The conditions are630ripe for a resounding success. We're very fortunate to have a631very strong panel of witnesses with us today who bring deep632expertise from across the advanced nuclear energy and633artificial intelligence sectors.634    I want to thank each of you for being here today, and I'm635looking very much forward to hearing your testimony.636    [The prepared statement of Chairman Babin follows:]637638    I want to thank our Energy Subcommittee Chairman and my639good friend, Mr. Weber, for holding this timely hearing on640powering data centers with advanced nuclear technologies.641    Artificial intelligence is advancing at a breakneck pace.642Every day, we see new groundbreaking developments and headlines643that highlight AI's expanding role in our economy, national644security, and daily lives.645    Fortunately, the United States currently leads the world in646the AI race--but that lead is not guaranteed. Both friends and647foes are moving swiftly to close the gap.648    At the core of AI innovation are data centers--the engines649that drive this technological revolution.650    These facilities require enormous and uninterrupted power.651And when it comes to reliability, only one option consistently652meets those high standards: nuclear.653    Industry standards for top-tier data centers, especially654those supporting AI workloads, require what's known as ``five-6559s'' uptime or 99.999 percent operational reliability. This656means that over the course of a year, the data center is657expected to be operational 99.999 percent of the time. That658equates to just 5.25 minutes of allowable downtime per year.659Few energy sources can meet that bar. In fact, nuclear power,660with its 92.5 percent capacity factor, stands alone in its661ability to provide the clean, constant baseload power these662systems demand. Plus, nuclear can provide much more energy per663square foot than its competing energy sources, allowing nuclear664to be sited nearer to customers and be less impactful on665overall land use.666    As this Administration has stated, nuclear energy will play667a central role in our energy future.668    Thanks to executive orders from President Trump and recent669actions from the Department of Energy (DOE), we are beginning670to cut through the red tape that has long stalled nuclear671progress--moving quickly, but safely to scale nuclear power672nationwide.673    To get to this point, the federal government has spent over674a decade significantly investing in advanced nuclear through675the DOE's Office of Nuclear Energy. These early-stage676investments were essential to getting first-of-a-kind677technologies off the ground. However, in today's budget-678constrained environment, that support must be seen as a679launchpad--not a permanent lifeline. It's time for the private680sector to step up.681    Fortunately, the market is ready. The nuclear renaissance682we anticipated in the early 2000s is now within reach. As the683demand for power increases, new nuclear companies have a684critical partner in large technology firms that are willing to685invest in and help deploy the fleet of nuclear reactors that686has long been promised.687    However, these tech companies are not only willing to pay a688premium to bring these new, first-of-a-kind power sources689online, but they have also shown an appetite for current690technology.691    Earlier this year, Microsoft and Constellation Energy--one692of our witnesses today--announced plans to reactivate the Crane693Clean Energy Center, marking a historic moment. Additionally,694just last week, Meta and Constellation announced a 20-year695agreement to keep the plant operational.696    This isn't just good energy policy--it's good economic697policy, creating well-paying jobs and generating millions in698tax revenue that would otherwise be lost.699    This momentum is spreading. Across the country--including700my home state of Texas--lawmakers are advancing pro-nuclear701policies. Universities are also seeing the benefits of702partnering with advanced nuclear. Texas A&M, through its RELLIS703campus, announced the creation of ``The Energy Proving Ground''704project. This project will involve Texas A&M assisting four705chosen SMR companies with permitting and providing a location706to demonstrate their reactors.707    Now is the time for the nuclear industry to take its next708step toward unlocking its full potential. The conditions are709ripe for resounding success.710    We are fortunate to have a strong panel of witnesses with711us today who bring deep expertise from across the advanced712nuclear energy and artificial intelligence sectors.713    I want to thank each of you for joining us today, and I714look forward to your testimony. With that, I yield back.715716    Chairman Babin. With that I yield back, Mr. Chairman.717    Chairman Weber. Thank you, Chairman Babin. I now recognize718the Ranking Member of the Full Committee for a statement.719    Ms. Lofgren. Well, thank you, Chairman Weber and Ranking720Member Ross, for holding this hearing today, and I want to721thank the witnesses for being here and for your expertise.722    We have a real challenge ahead of us. We've seen incredible723strides in the capabilities of artificial intelligence over the724last few years and its application to research, industry,725agriculture, and other sectors of our economy. However, we also726know that training and running AI models can consume enormous727amounts of energy. And while certainly not the only solution,728advanced nuclear technologies are quite promising in their729potential to meet these expected needs.730    And over the past decade in particular, this Committee has731developed and enacted bipartisan, comprehensive legislation to732explore in advance of this research--resource. In fact, the733last bill we had before this Committee on next-gen nuclear was734adopted unanimously, and this is another reason why I am so735disappointed with the Administration's budget proposal for7362026.737    The budget also states that it--and this is a quote--738``unleashes America's energy dominance through funding for739nuclear energy.'' But it would cut support for DOE's Office of740Nuclear Energy by 21 percent, and slash funding for the741flagship Advanced Reactor Demonstration Program by 51 percent.742    The budget also states that it is ``prioritizing fusion743research,'' but it proposes to cut fusion research by 6744percent. Now, this may be old fashioned, but I think words745actually should mean something, especially when they come from746the U.S. Government, and these words do not reflect the actual747proposal.748    Now, some may note that the budget request does include749support for loan guarantees for nuclear technologies, and750that's all well and good until you look at the rescission751reconciliation bill that the Republicans in the House passed,752which, of course, the President endorsed. If enacted, that bill753would eliminate more than four times as much support for DOE's754loan guarantee program as this budget proposal would provide.755It would cut tax incentives that industries told us are756critical, in tandem with a robust Federal loan program, to757enabling the widespread deployment of new nuclear power plants.758    So I look forward to discussing these stark contradictions759and other challenges to our clean energy future with this760really excellent panel of witnesses. And with that I want to761thank you all for being here. And so we can get directly to762your testimony, I yield back the balance of my time.763    [The prepared statement of Ms. Lofgren follows:]764765    Good morning and thank you, Chairman Weber and Ranking766Member Ross, for holding this hearing today. And thank you to767the witnesses for being here this morning.768    We have a real challenge ahead of us. We have seen769incredible strides in the capabilities of artificial770intelligence over the last few years and its applications to771research, industry, agriculture, and various other sectors of772our economy.773    However, we also know that training and running AI models774can consume enormous amounts of energy. While certainly not the775only solution, advanced nuclear technologies are quite776promising in their potential to meet these expected needs. And777over the past decade in particular, this Committee has778developed and enacted bipartisan, comprehensive legislation to779explore and advance this resource.780    This is another reason why I am frankly so disappointed781with the Administration's budget proposal for 2026. While the782budget plainly states that it ``unleashes America's energy783dominance through funding for nuclear energy,'' it would cut784support for DOE's Office of Nuclear Energy by 21%, and slash785funding for its flagship Advanced Reactor Demonstration Program786by 51%.787    The budget also states that it is ``prioritizing fusion788research'' while proposing to cut fusion research by 6%. Call789me old fashioned, but I think words actually need to mean790something, especially when they come from the U.S. government,791and these words are sadly hollow.792    Now some may note that the budget request does include793support for loan guarantees for nuclear technologies, and794that's all well and good until you look at the House-passed795Republican reconciliation bill, which of course the President796endorsed. If enacted, that bill would eliminate more than 4797times as much support for DOE's loan guarantee program as this798budget proposal would provide. And it would gut tax incentives799that industry has informed us are absolutely critical, in800tandem with a robust federal loan program, to enabling the801widespread deployment of new nuclear power plants.802    So, I look forward to discussing these stark contradictions803and other challenges to our clean energy future with this804excellent panel of witnesses. With that, I thank you all again805for being here, and yield back the balance of my time.806807    Chairman Weber. The Ranking Member of the Full Committee808yields back. I now recognize Deborah Ross to the right of me.809    Ms. Ross. OK, thank you, Mr. Chairman. I ask unanimous810consent (UC) that Mr. Beyer from Virginia be permitted to811attend this hearing and after all Committee Members have had812their opportunity to ask questions of the witness.813    Chairman Weber. Without objection.814    I do want to remind something I left out earlier. Members815are reminded that this Committee's practice is to submit816letters and other items for the record to the Committee before817seeking unanimous consent to insert such items into the hearing818record. This allows the Committee to ensure that such items819meet the rules of decorum, and verify their length and820provenance that they do not contain sensitive, proprietary, or821controlled information. The Chair reserves the right to object822to any UC request to insert items in the record that are not823provided in advance.824    So let me introduce our witnesses.825    Our first witness today is Mr. Pat Schweiger, the Chief826Technology Officer (CTO) of Oklo.827    Our next witness, Ms. Kathleen Barron--am I saying that,828Ms. Kathleen?829    Ms. Barron. Barron.830    Chairman Weber. I can do this. Barron, OK, Executive Vice831President and Chief Strategy and Growth Officer at832Constellation Energy.833    Our final is Dr. Jeremy Renshaw--thank you for the easy834name to say--Executive Director of AI and Quantum at EPRI.835    I now recognize Mr. Schweiger for 5 minutes to present his836testimony.837838                TESTIMONY OF MR. PAT SCHWEIGER,839840                 CHIEF TECHNOLOGY OFFICER, OKLO841842    Mr. Schweiger. Good morning, Chairman Weber, Ranking Member843Ross, and Members of the Committee. Thank you for the844opportunity to testify and for holding this important hearing.845My name is Pat Schweiger, and I am the CTO at Oklo, an advanced846nuclear technology and fuel recycling company. Prior to Oklo I847worked 21 years at the Fast Flux Test Facility, an848internationally recognized premier sodium fast reactor where we849tested advanced fuels and materials for fusion and fission. We850set performance records that no other nation has been able to851achieve, even to this day.852    Oklo is developing fast fission power plants known as853Aurora powerhouses to provide clean, reliable, and affordable854energy at scale. Oklo is at the forefront of transforming the855technological basis and business model associated with nuclear856power in America. In our build, own, and operate business model857we plan to sell power in the form of electricity and heat858directly to customers, where we believe we can allow for fast859track customer adoption. Most importantly, we are860commercializing fast reactor technology pioneered by the U.S.861Department of Energy over 50 years ago, and have a site use862permit from DOE for our commercial reactor at Idaho National863Laboratory (INL).864    AI has triggered a Sputnik moment, accelerating the demand865for dependable domestic power. According to Goldman Sachs, AI866data centers will have a significant contribution to power867demand growth, driving a 160 percent increase in power demand868through 2030. Our partnership with Equinix was the first869commercial advanced nuclear energy deal in the data center870industry that included an investment from a data center company871to a nuclear company.872    Energy is the foundation upon which America's AI future873depends. Oklo represents a new approach to leveraging the874benefits of mature nuclear power generation to meet the growing875energy demands associated with emerging AI applications. The876market has rewarded this approach with a pipeline of over 14877gigawatts of commitments from prospective customers, and most878recently through a 12 gigawatt master power agreement with AI879and data center provider, Switch, which Randy mentioned, one of880the largest corporate clean power agreements in history.881    Oklo's powerhouse builds on America's investment in882cutting-edge nuclear technology in the first atomic age. Right883now there are no fast reactors operating in the U.S., but Oklo884will change that, leveraging a legacy that blossomed with885research and test reactors such as the Fast Flux Test Facility886and Experimental Breeder Reactor 2, which was a fast reactor887that ran at INL for 30 years and a capacity of 20 megawatts of888electric power.889    Oklo's reactors are based on this proven liquid metal890cooled sodium fast reactor technology. The reactor is self-891stabilizing, self-controlling, and cooled by natural forces.892This means the plant is walk-away safe, and can be sited in893closer proximity to populated areas, crucial locations for data894centers, and other AI infrastructure, as Ms. Ross noted, as895well.896    Additionally, fast reactors can derive energy from spent897nuclear fuel. Thanks to U.S. innovation, spent nuclear fuel can898be recycled, and is being done at our national labs today.899Idaho National Lab is producing spent fuel from EBR-II into900HALEU, and Argonne National Lab is advancing the technology901further with support through ARPA-E. Fast reactors are ready to902be commercialized and poised to meet this moment for AI.903    To meet the needs of this critical moment in our country, I904want to offer the Committee a few policy changes to accelerate905advanced nuclear deployment.906    No. 1, unlock an abundance of nuclear fuel. Congress should907continue to push DOE to accelerate its support of the domestic908fuel supply chain and HALEU production, and to think creatively909about new ways to enhance the domestic fuel supply, including910the accelerated processing of DOE spent fuel into HALEU.911    No. 2, continue investment in next-generation research.912U.S. Government research is driving American nuclear913innovation. Research programs in the fuel cycle, commercial914fuel recycling, and next-generation core technologies are915necessary to compete globally.916    And then, finally, No. 3: modernize regulations around917technologies with decades of proven safety. Congress should918rethink how we regulate inherently safe, proven nuclear919technologies, from advanced reactors to commercial fuel920recycling to waste management, so that American nuclear plants921can serve energy needs for AI civilian, DOE, and military922installations. It's inevitable that advanced nuclear reactors923will be part of the energy solution to ensure U.S. leadership924in AI. Our groundbreaking approach is redefining nuclear925energy, making it safer, faster to deploy, and more cost926effective than ever before.927    Oklo is ready to work with the Committee and Members of the928House to ensure the success of both the nuclear and AI929industries. I look forward to today's discussion.930    [The prepared statement of Mr. Schweiger follows:]931    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]932933    Chairman Weber. Thank you, Mr. Schweiger. You have a very934interesting quote, which I love, when you say AI has triggered935a Sputnik moment. Is that a--I don't mean to put you on the936spot, but is that original with you?937    Mr. Schweiger. Well, not to me personally, but original to938who wrote--help me write this.939    [Laughter.]940    Chairman Weber. Well, that's is a very, very good point.941    Ms. Barron, I am going to yield 5 minutes to you. Thank942you.943944              TESTIMONY OF MS. KATHLEEN L. BARRON,945946          EXECUTIVE VICE PRESIDENT AND CHIEF STRATEGY947948            AND GROWTH OFFICER, CONSTELLATION ENERGY949950    Ms. Barron. Good morning, Full Committee Chairman Babin,951Ranking Member Lofgren, Chairman Weber, Ranking Member Ross,952and Members of the Subcommittee. Thank you for the opportunity953to appear before you to discuss the role of nuclear energy in954powering America's artificial intelligence infrastructure.955    Constellation is the largest owner and operator of956commercial nuclear plants in the United States. We operate 21957reactors in Illinois, Maryland, New York, and Pennsylvania, and958we have an ownership interest in four additional reactors in959New Jersey and Texas. But we also have a diverse portfolio of960power generation resources. All in, we make 32 gigawatts of961electricity, which is equivalent to powering 16 million homes962and businesses. I'd like to make three points from my testimony963today.964    First, there should be no debate. America must win the race965for AI supremacy. And to do that we need to assure timely power966supply for AI infrastructure like data centers, while at the967same time securing reliable, affordable, and clean power for968all customers.969    Second, the Nation's existing nuclear power fleet can help970meet the near-term need for power by extending the operating971life of existing reactors, by increasing the output of the972existing fleet of plants, and by, as has been mentioned,973restarting previously closed reactors that are capable of974resuming operation.975    And third, advanced reactors can add enormous quantities of976reliable, affordable, and clean energy to the grid in the977longer term. The most logical place, in our view, to think978about siting new nuclear plants is at sites that are already979hosting nuclear reactors, and that's because these sites have980already been proven to meet environmental and safety-related981regulatory requirements, and have critical existing cooling982water, rail, and electric infrastructure to host these new983reactors.984    You know, and the biggest thing is really these incredible985communities that host the existing fleet, which is comprised of986hundreds of workers and their families, are supportive of this987opportunity to add new nuclear. And they also have land988available where we could host sites like data centers that can989collocate and minimize the need for additional electrical990infrastructure and transmission.991    As has been mentioned by the Full Committee chair,992Constellation has been working to enable data center993development in the places where we operate. Last September we994announced that we will restart unit one, which is the undamaged995reactor at Three Mile Island. It was one of our best performing996reactors when it closed prematurely before the end of its997licensed life in 2019 due to economic factors partly caused by998poor policy choices. But it will resume operation as the Crane999Clean Energy Center as part of a 20-year power purchase1000agreement with Microsoft. When that plant is returned to1001service, it will provide 835 megawatts of power to the PJM grid1002for use at Microsoft facilities across the PJM region.1003    And then earlier this month, we announced another 20-year1004power purchase agreement with Meta, this time for the output of10051,100 megawatts at our Clinton Clean Energy Center in central1006Illinois, to support Meta's facilities in the Midwest region1007beginning in 2027. This agreement will allow us to relicense1008the Clinton station for another 20 years, and allow it to1009operate until at least 2047. And it also calls for us to uprate1010the plant, or increase its output by an additional 30 megawatts1011of power. It also allows us to evaluate strategies to extend1012the plant's existing early site permit at the NRC (Nuclear1013Regulatory Commission), or perhaps to seek a new construction1014permit from the NRC to pursue development of an advanced1015reactor at that site.1016    These agreements ensure that the Crane and Clinton1017facilities will remain on the grid for at least two decades to1018support economic growth and to power America's artificial1019intelligence infrastructure. But to ensure all of the reactors1020stay online and expand to meet this increased demand,1021supportive policies are critical.1022    President Trump's recent executive orders direct that the1023Department of Energy shall prioritize work with the nuclear1024energy industry to facilitate 5 gigawatts of power uprates to1025existing nuclear reactors, and have 10 new large reactors with1026complete designs under construction by 2030. These are1027appropriately ambitious goals. But in order to achieve them we1028recommend first that Congress continue the Section 45U1029production tax credit for existing nuclear plants, as well as1030the Section 45 and 48 technology neutral tax credits for1031nuclear generation, which is consistent with comments from the1032Administration and the recently passed House reconciliation1033measure.1034    Second, Congress should retain funding for the Department1035of Energy's Loan Program's Office for Nuclear Investment.1036    Third, Congress should continue to support Department of1037Energy programs within the Office of Nuclear Energy to support1038research, development, and demonstration activities related to1039advanced nuclear power, including the Advanced Reactor1040Demonstration Project.1041    And finally, Federal agencies like the Federal Energy1042Regulatory Commission should remove barriers that prevent data1043centers from accessing and using available sources of energy.1044FERC has been debating for over a year the rules for data1045centers to collocate with power plants. As the President has1046recognized, collocation is--which is when the data center is1047sited right at or near the power plant, as Ranking Member Ross1048mentioned--enables development on a quick basis. And that's1049because it minimizes the need for new transmission lines to1050deliver power over long distances, which also lowers costs for1051both the data centers and all customers.1052    Data center projects should be permitted to access the grid1053using the configuration that makes the most for that facility1054and in that location, and should not be slowed down by a lack1055of clear Federal rules.1056    So thank you again for the opportunity to appear before you1057today, and I look forward to your questions.1058    [The prepared statement of Ms. Barron follows:]1059    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]10601061    Chairman Weber. Thank you, ma'am.1062    Dr. Renshaw, you've got a hard act to follow. You're1063recognized for 5 minutes.10641065                TESTIMONY OF DR. JEREMY RENSHAW,10661067                EXECUTIVE DIRECTOR AI & QUANTUM,10681069            ELECTRIC POWER RESEARCH INSTITUTE (EPRI)10701071    Dr. Renshaw. Thank you. Chairmen Weber, Babin, and1072Ranking----1073    [Audio malfunction.]1074    Chairman Weber. Dr. Renshaw, I'm sorry. Turn your mike on.1075    Dr. Renshaw. I'm sorry. Would you like me to start over?1076    Chairman Weber. In a word? Heck, no.1077    [Laughter.]1078    Dr. Renshaw. All right. EPRI has worked with AI for1079decades, observing rapid growth in computational needs1080accelerated by recent breakthroughs in generative AI. Advances1081in GPUs (graphics processing units) have improved compute1082efficiency, but these gains have been outpaced by the1083increasing size of AI models. AI is redefining how knowledge is1084created.1085    The Industrial Revolution used machines to turn raw1086materials into goods more efficiently, and now the AI1087revolution is using data to turn--to accelerate productivity1088and discovery across all industries. Thus, AI and energy1089industries have become intertwined, with more energy needed to1090support AI, and AI enabling more efficient and productive1091energy systems. Therefore, EPRI launched the Open Power AI1092Consortium to build a collaborative ecosystem between the1093energy and technology industries to maximize benefits to1094stakeholders and the public.1095    With over 100 organizations engaged, the consortium will1096focus on building more efficient and performant AI models to1097achieve better results with less compute and energy needs. For1098example, accelerating interconnection queues which have led to1099bottlenecks in connecting new generation, transmission, and1100distribution infrastructure to the grid.1101    Predicting the future energy needs of AI is challenging.1102Recent work by EPRI and others points to significant growth and1103uncertainty in the future power consumption of data centers1104from AI, which are influenced by model size, volume of training1105data, inference loads, adding reasoning capabilities, hardware1106efficiency, and more. Moreover, future innovations could alter1107the trajectory of energy needs including novel chip designs,1108advanced computing and model architectures, emerging compute1109modalities such as quantum computing, and software1110optimization.1111    AI is delivering value globally, driving growth in data1112center utilization and energy use. AI-specific data centers1113consume up to five times more energy than traditional data1114centers. The International Energy Agency predicts that global1115data center energy use in 2030 will double to 945 terawatt1116hours, more than Japan's total electricity use today. AI data1117centers may have highly variable loads with spikes in energy1118demand, compared to traditional flat loads from data centers.1119This variability presents a challenge and an opportunity for1120the grid. While data centers represent substantial new loads,1121they also offer opportunities such as workload flexibility and1122utilizing backup generators as a dispatchable grid resource.1123    EPRI launched the DC Flex Initiative to explore how data1124centers can provide these grid services supporting utilities,1125operators, and consumers alike. Some perceive data centers as1126growing unsustainably and straining the grid. The Open Power AI1127Consortium and DC Flex Initiative can change that perspective1128and utilize data centers to accelerate productivity, knowledge1129generation, and innovation across all sectors while growing1130responsibly. Advanced forms of energy generation may support1131data center energy needs.1132    Advanced nuclear is one of several options for powering the1133next generation of AI infrastructure. Having worked in the1134nuclear industry for over 15 years, I will discuss the benefits1135and limitations of nuclear to support growing energy demands.1136No energy source is perfect, and all have benefits and1137limitations, and a robust energy mix combines multiple sources1138to improve overall system performance, reliability, and reduced1139risk.1140    Nuclear plants provide safe, reliable, and carbon-free1141baseload power. Nuclear power has a track record of being one1142of the safest forms of energy generation over the last several1143decades. Advanced nuclear reactors offer additional capability1144for flexible operation, improved safety, efficiency, and a1145range of fuel sources, including spent fuel, supporting high-1146intensity variable loads like AI data centers, and reducing1147waste. However, fuel and irradiated materials must still be1148managed. As with any technology, first-of-a-kind1149implementations include risks such as delays and cost overruns.1150Other clean, reliable generation sources can augment overall1151energy system performance.1152    In conclusion, AI is transforming our society and its1153energy demands are growing rapidly. Improvements in chip1154efficiencies reduce energy use while larger models and1155increased utilization increase energy use. AI is accelerating1156productivity and knowledge generation across all industries1157today, and is poised to continue. Meeting data center energy1158needs is a growing challenge, with many solutions being1159evaluated. While energy intensive, data centers can also be a1160part of the solution via flexible operation and grid1161integration.1162    While there is no perfect energy source, advanced nuclear1163is among the options that offer safe, reliable, flexible, and1164clean power to meet future needs. Nuclear technologies can play1165an important role to support the growing needs of AI and the1166benefits it can provide to society.1167    [The prepared statement of Dr. Renshaw follows:]1168    [GRAPHICS NOT AVAILABLE IN TIFF FORMAT]11691170    Chairman Weber. I thank the witnesses for their testimony.1171I now recognize myself for 5 minutes.1172    Mr. Schweiger, unlike regulatory bodies such as the NRC,1173Nuclear Regulatory Commission, the Department of Energy is an1174industry-facing institution advancing the commercialization of1175new technologies. I think we'd all agree with that.1176    Through DOE's programs, Oklo has secured EBR-II fuel, and1177selected Idaho National Laboratory for its Aurora reactor. So,1178in your opinion, how important is DOE to Oklo's success and its1179ability to provide those 12 gigawatts of power to hyperscalers1180like Switch?1181    Mr. Schweiger. Thank you for the question and also for your1182work on the advanced program. We appreciate that, the industry.1183    I would say I worked with the DOE about 45 years off and1184on, and the Fast Flux Test Facility was in that DOE program.1185EBR-II and the recycled fuel that we're going to receive, I1186think, is essential. When my associates in the industry say,1187well, what's Oklo going to do for fuel, I go, well, it's1188handled. And I think it--I don't know this for darn sure, but1189I'm pretty sure it's the only SMR that's got a fuel supply1190ready and available--or, you know, in the process of being made1191available through DOE.1192    So that program gets us to where we want to be faster, and1193that's crucial right now.1194    Chairman Weber. Especially in the development of AI.1195    Some of my colleagues on the other side of the aisle may1196complain that the President's Fiscal Year 2026 budget, as well1197as DOE's reorganization, will hurt companies like Oklo. Can you1198describe your interactions with the Trump Administration and1199any of their actions that slowed the development of Aurora?1200    Mr. Schweiger. Mm-hmm. So as CTO, I'm not as tightly1201coupled with the business side, as you can imagine. But the1202Oklo business plan is less reliant on the Federal Government.1203So what Oklo is trying to do is fund, go build power plants,1204and then sell that power. And we're well on track for that with1205private funding.1206    So Oklo uniquely is not going to be much affected by the1207government policies like with the Federal loan program. So1208that's what I can comment to. I don't know the inner workings1209of the rest of it.1210    Chairman Weber. OK, very few of us do. I thank you for1211that.1212    Ms. Barron, I'm going to come to you. Over the last decade,1213non-regulated utilities have been at the forefront of1214displaying next-generation nuclear reactors. Southern Company1215completed the construction, as we all know, of two AP1000s at1216Vogtle, and the TVA, Tennessee Valley Authority, recently1217announced that it submitted a construction permit for GE1218Vernova's BWRX-300 at its Clinch River site.1219    Can you explain for our benefit why merchant markets are1220lagging in deploying next-generation nuclear reactors, compared1221to these non-regulated entities? What's inhibiting, for1222example, your company from constructing a new reactor?1223    I will yield the time to you.1224    Ms. Barron. Thank you for the question, Chairman.1225    So as you pointed out, the--operates where--the markets1226where Constellation operates are competitive markets. That1227means that, moment to moment, a system operator chooses which1228asset to run based on cost. And if you get picked because1229you're competitive, you run. And if you aren't, you don't. And1230customers, you know, get the benefit of that. Over the years1231customers have seen much lower costs in the competitive markets1232because of that dynamic. There is no guaranteed rate recovery1233for anyone that operates in a competitive market.1234    So you contrast that to the monopoly markets, where1235regulators will make a decision to invest in a technology. And1236in a guarantee rate recovery, that is the explanation for why1237you had seen Vogtle move forward in Georgia and in VC Summers's1238case in South Carolina.1239    I think what's changed is that there are now corporates1240that are looking to help fund investment in new technology. We1241have supportive Federal policy that is helping many reactor1242developers, including Oklo, come to the market. And we have an1243interest across the financial sector, the OEMs (original1244equipment manufacturers), and then ultimately the operators to1245work together to try to figure out how to bring these projects1246to bear.1247    So I wouldn't look to the past as an indicator of the1248future. I think you'll see some of these developments in1249competitive markets, as well.1250    Chairman Weber. I wish I could train my wife not to look to1251my past as something about the future.1252    [Laughter.]1253    Chairman Weber. But I want to come back to you. How many1254customers of y'all's would you say that that affects when1255you're trying to make sure the prices are the best? How many1256customers do you all serve?1257    Ms. Barron. So we make enough--we're a wholesale producer1258of electricity, so we make enough electricity to put out into1259the grid to serve 16 million homes and businesses. On our1260competitive--the competitive side of our business, where we1261sell electricity in places, where that's permitted, i.e. the1262non-monopoly markets, we serve about three-quarters of the1263Fortune 100. So we have both large commercial industrial1264customers and then residential customers that we rely on. And1265we have to compete to serve those customers by providing the1266best price and the best product.1267    Chairman Weber. Right. And the reason I ask that is for the1268benefit of everybody watching, you all are trying to do the1269best thing for the most people. We appreciate that.1270    I'm going to jump now to this question. With technology1271companies and hyperscalers taking an active role in procuring1272power for AI--you all have got to be watching that--do you1273believe that this paradigm shift will empower companies like--1274my writing--they wrote this ``like yours,'' but in Texas we say1275like y'all's, OK? Like y'all's. Do you believe that that will1276affect y'all's ability to build a new reactor into the market?1277    Ms. Barron. You know, I think, from our perspective, we1278think you should do the cheapest thing first. And the cheapest1279thing is to ensure that the existing fleet remains in1280operation.1281    You know, unlike other sources of technology, nuclear needs1282to be licensed by the Federal regulator, which does a great1283job. And there's a finite licensed life to the existing1284reactors. So investing in the plants, asking for a second1285license extension, allowing the plants to run another 20 years,1286these plants can run well past mid-century into the 2070s.1287    Second, we can uprate them or do modifications at the site1288to create more output from the same physical plant, and those1289are investments that we have underway. And all-in across just1290our fleet, if we were to do all the remaining uprates available1291we could make another 1,000 megawatts, so equivalent to a whole1292new reactor.1293    Chairman Weber. Well, thank you. I've well over-stated my1294time, so I appreciate your diligence. And I'm now going to1295yield at least 5 minutes to the Ranking Member.1296    Ms. Ross. Thank you, Chairman Weber. I'm going to take a1297little bit at the end, but not to quiz our witnesses.1298    Thank you all for your very insightful and comprehensive1299testimony.1300    Ms. Barron, I'm going to go--talk a little bit about the1301monopoly markets, because North Carolina is a monopoly market.1302And while meeting the significant energy demands of AI1303infrastructure, we also worry about our ratepayers in the1304monopoly market. And as we discussed, in South Carolina the1305ratepayers really were on the hook.1306    In North Carolina there's a debate within our legislature1307about whether to allow construction work in progress, which we1308don't allow for nuclear. And so I worry that ratepayers can end1309up with higher energy prices like what happened in South1310Carolina when the project is failed or there are long delays.1311    And I support the need for more nuclear, but in that1312monopoly market it becomes difficult with the ratepayers. It1313also becomes difficult when the data centers want to have their1314own sources of energy because the monopoly market doesn't1315really like that very much. And I had that experience even on a1316military installation when I was in--when I was practicing law1317and trying to help get solar panels at Fort Bragg, of all1318things.1319    So can you describe how utilities can protect ratepayers1320from these unnecessary costs while also advancing other1321projects like nuclear?1322    Ms. Barron. Thank you.1323    Chairman Weber. Mike on.1324    Ms. Ross. Mike on.1325    Ms. Barron. Thank you for the question. And as I mentioned,1326we do operate in competitive markets. So I'm a little bit1327outside of my lane.1328    But I do think it's fair to acknowledge that all first-of-1329a-kind technologies have challenges with remaining on time and1330on budget, and we've seen that on the East Coast with a number1331of programs supporting offshore wind up and down the Mid-1332Atlantic and into the northeast, where, you know, unexpected1333cost increases have led to having--States having to renegotiate1334contracts, and the challenge that you identified being front1335and center, because these programs are funded by customers.1336    That being said, these are long-lived assets. I mean, these1337stations, according to the NRC, can run for 80 years, maybe1338longer, but they do take a long time to build. And during that1339period of construction--I agree it is a challenge to how you1340can you can manage those costs and make sure that the risk is1341shared and is not exclusively borne by customers. In our case,1342in a competitive market, we would look to a customer to help1343support the project during development, and ultimately1344guarantee the offtake so that we can share that risk and not1345create the situation that you mentioned that you're facing in1346North Carolina.1347    Ms. Ross. Thank you. My next question is--sorry, my next--1348thank you, Mr. Chairman--my next question is for Dr. Renshaw.1349    At the end of your testimony you talked about an energy1350mix, and how we can maybe use more intermittent resources in1351conjunction with baseload power, and actually even sell energy1352back to the grid, be--have this dispatchable resource that may1353be backup, and helping other consumers of energy. Are there any1354examples going on right now in the country where you can tell1355us this--this really works, it's a great model?1356    Dr. Renshaw. Yes. So first, if you'll indulge me for a1357second, we can also cover where it doesn't work. And we've seen1358this in the past when Russia invaded Ukraine, gas prices went1359through the roof, and many European utilities were in a very1360difficult situation, losing millions to hundreds of millions of1361dollars per day based on the increase in gas prices because1362they were overly reliant on one source. That's why a mix of1363energy sources is important that are clean, safe, affordable,1364and environmentally responsible.1365    One thing that is exciting is EPRI just announced this1366morning, as part of the DC Flex Initiative, three test sites1367that we will be starting from locations around the world to1368evaluate data center flexibility in terms of using those data1369centers, backup generators potentially powered by clean fuels,1370to operate flexibly and provide power back to the grid in a way1371that we can either shift the time or location of workloads or1372use those backup generators.1373    Ms. Ross. So having this backup as dispatchable and also1374being able to do load shifting could be a really good model?1375Great.1376    I'm going to use my remaining 15 seconds to thank Joseph,1377right here, for his amazing service to the SST Committee. He is1378going back to North Carolina to--to work in the area of--area1379of nuclear because, it's really so important in North Carolina.1380And so I'd like the Committee to give him a round of applause.1381    [Applause.]1382    Ms. Ross. And I yield back.1383    Chairman Weber. OK. The chair now recognizes the Full1384Committee, Dr. Babin, for 5 minutes.1385    Chairman Babin. Thank you very much, Mr. Chairman.1386    Mr. Schweiger and Ms. Barron, several States, including my1387home State of Texas, have passed legislation to help attract or1388develop the nuclear industry there. Texas House Bill number 14,1389the Texas Advanced Nuclear Deployment Act, has been sent to the1390Governor's desk to be signed into law. This bill creates a1391State office, the Texas Advanced Nuclear Energy Office, to1392identify regulatory and financial barriers, promote public1393education, and support the growth of a nuclear energy supply1394chain. It also establishes the largest state-level grant1395program in the Nation to develop nuclear projects.1396    How do actions like this support Federal investments that1397allow this sector to advance from DOE projects to actual1398electrons on the grid?1399    Ms. Barron.1400    Ms. Barron. Thank you for the question. And, I mean, I1401think the answer is leadership matters. I think the State of1402Texas has spent a lot of time focusing on this, starting at the1403commission and the task force that was formed, Commissioner1404Glotfelty's report. And then, obviously, the work the1405legislature did to enact the bill that you referenced. It makes1406a big difference, and it sends a signal that this is important,1407and that the State supports it.1408    And we are seeing that leadership across a number of our1409States. The Governor of New York has done the same thing,1410launched a process to figure out how the State can encourage1411new reactors in the State. Maryland passed a bill that supports1412the addition of new reactors, as well. So I think those are all1413tremendous signs.1414    Chairman Babin. OK.1415    Ms. Barron. To get to the nub of your question, though, you1416know, to get these reactors actually onto the grid is a1417challenge. The industry is going to need to meet this moment.1418And we have a lot of leadership here at the table to--to talk1419about that further. But I agree with the premise of your1420question that these actions at the State level make--make a1421very big difference to the industry.1422    Chairman Babin. OK. And then I'm going to ask Mr.1423Schweiger.1424    I'm going to ask you a little bit different. Oklo has1425championed the use of milestone-based contracts to support the1426liftoff of advanced reactors similar to NASA (National1427Aeronautics and Space Administration's) Commercial Orbital1428Transportation Services, or COTS, C-O-T-S, which led developed1429commercial cargo delivery capabilities to the International1430Space Station. In that instance NASA required skin in the game1431from the contractors equal to 50 percent of the development1432costs, and spread the development risk across multiple1433contractors.1434    What are the benefits of a milestone approach, and should1435DOE continue to use this form of contracting in the future, as1436they did recently in their request for proposals for the Gen1437III+ demonstration project?1438    And what conditions should be included in milestone-based1439contracts to ensure scheduled discipline?1440    Mr. Schweiger. OK, thank you----1441    Chairman Babin. That's a two-part question.1442    Mr. Schweiger. Yes, thank you for the question.1443    I'd like to add to your first question that education is so1444crucial. If you think back to post-Fukushima, people's desire1445to use nuclear was pretty low, and they just didn't understand.1446    Chairman Babin. Right.1447    Mr. Schweiger. OK. Then, for the milestone-based program, I1448think there's inherently value in it. The Oklo approach is1449going to be less tuned to that because of our business model,1450which is to build the plants and then sell the electricity.1451    When you look across the nuclear--all the--the Gen IV1452plants that are trying to build new reactors and all those1453initiatives, milestone-based--what I like about it is you have1454to perform to get funding, instead of just getting a tranche of1455money that may or may not produce something. So the milestone,1456presumably, would have performance milestones in there. And1457then, when achieved, then more money can be released.1458    Chairman Babin. Right.1459    Mr. Schweiger. So, yes. So as far as what should be in the1460milestones, I'm not----1461    Chairman Babin. Yes, what conditions.1462    Mr. Schweiger. Yes, conditions. That's going to take a1463little bit of thought, but----1464    Chairman Babin. Well, I don't have but 37 seconds left.1465    [Laughter.]1466    Mr. Schweiger. I know. There we have a problem.1467    I think conditions for milestone-based would be do you have1468a technology that's viable? You know, the person seeking1469milestone support, is their technology actually viable? So, you1470know, some sort of evidence. Are you at a technology readiness1471level of four, five, six, somewhere in there?1472    And then, when you've established credible technology, then1473marking the progress of that. And I'm--I'm partial to that1474technology readiness assessment process that NASA pioneered and1475DOD has been using--and DOE.1476    Chairman Babin. Yes. OK, thank you.1477    And I yield back, Mr. Chairman. I have another question,1478but we'll have to submit that for the record.1479    Chairman Weber. Thank you, sir. The Chairman recognizes the1480Ranking Member of the Full Committee, Zoe Lofgren of1481California.1482    Ms. Lofgren. Well, thank you, Mr.----1483    Chairman Weber. For as much time as she may consume.1484    Ms. Lofgren. Well, thank you, Mr. Chairman.1485    You know, when we look at the international landscape, and1486particularly at China, there's--we're seeing massive levels of1487investment toward AI infrastructure, as well as nuclear energy.1488And I think they're building more nuclear than anyone else in1489the world. Meanwhile, we're debating here whether to limit1490programs that provided research and loans and the like.1491    How would each of you rate our ability to compete with1492China's massive nuclear expansion, particularly in developing1493next-generation reactor technologies?1494    And specifically, do any of you think it's a good idea to1495cut support for the Advanced Reactor Demonstration Program by1496more than half, as the Administration has proposed in its1497budget resolution?1498    Whoever wants to go first.1499    Ms. Barron. I can just make a few comments. I have some1500understanding of how the development is--is occurring in China.1501And, you know, I think when you have a centralized authority1502that is in charge, and you have construction crew A, B, C, D,1503E, and you can sort of dispatch them around the country, you1504can move faster. We have a different model here.1505    We have bifurcated authority between the Federal Government1506and the States over energy policy, and that, you know, keeps1507those of us up here in a job for a long period of time. But it1508does mean that it's a bit more--more complicated. So--but it1509allows, you know, involvement and--and appropriate input across1510different levels of government. And that's important.1511    So I think our challenge, as I mentioned a moment ago, is--1512is to sort of meet this moment now, and put all of our effort1513into trying to move forward as fast as we possibly can, and we1514have seen a lot of support for that of late, which I think is1515important.1516    Dr. Renshaw. If I can add to that, I would say China is1517definitely moving fast. They have the infrastructure in place,1518manufacturing capabilities that have accelerated their ability1519to perform. Currently they are on pace to build reactors in1520about 52 months, so just over 4 years, and they're doing that1521on time and on budget, meaning that they are growing in1522credibility and trust with the people who are ordering those1523plants. So that's a credibility and trust that would be1524important to have in other regions of the world, to be able to1525say this is how much a reactor will cost to build, and this is1526how long it will take.1527    Ms. Lofgren. You know, it just seems to me--I--the Chairman1528mentioned Fukushima, which made people around the world1529nervous. And the legacy systems are different than the next-1530generation system, which has broad support. And I'm thinking1531about my own State. Near Morro Bay there's a legacy nuclear1532plant that was going to be decommissioned because, you know,1533building a dam or a nuclear plant is a way to find an1534earthquake fault, and they found new earthquake faults near the1535facility. We're keeping it open because of the energy needs,1536but people are uneasy about it because it's a legacy system,1537and we don't have that unease about the next generation.1538    So what about the Advanced Reactor Demonstration Program1539and the reduction that's being proposed? Does that make a1540difference for our future?1541    Dr. Renshaw, do you have an opinion?1542    Dr. Renshaw. Well, EPRI doesn't comment on government1543policy.1544    Certainly, investments in research and development can help1545to accelerate the processes that we have, as well as the1546technologies. So I would say that is--if we are investing1547correctly, then it helps to accelerate all forms of research1548and development, whether it's nuclear or otherwise.1549    Ms. Lofgren. One of the things--and I'm glad that our1550colleague, Mr. Beyer, is here, he's the co-chair of the Fusion1551Caucus--but, you know, we're skating toward where the puck is1552going to be here. And both in terms of new energy sources, but1553also the energy use, there is some in the AI space who believe1554that the power consumption is actually going to go down as1555quantum comes into play and, using different algorithms, that1556the energy issue is going to be different than it is today. Do1557you have a view on that, Dr. Renshaw?1558    Dr. Renshaw. Yes, I would say--I would refer back to my1559testimony that the future energy needs are very uncertain.1560Certainly, quantum computing is a technology that holds1561significant promise in terms of accelerating certain types of1562computing problems. AI may be one of those problems, as well as1563optimization, search materials development, and so forth. So1564there is the potential that future quantum computing modalities1565would help to significantly reduce the--the amount of energy1566that's required to train and test models.1567    But I would also point out Jevons Paradox, which1568interestingly came out of the coal industry, where the1569increases in efficiency of using a technology often results in1570the expansion of usage of that technology. So if we can make1571the use of quantum computing help AI training and inference,1572then there's the potential for massively increased usage of AI1573and other technologies.1574    Ms. Lofgren. Thank you, Mr. Chair. My time has expired, so1575I yield back.1576    Chairman Weber. Thank you, ma'am. The Chair now recognizes1577Lieutenant Colonel Biggs from South Carolina for at least 51578minutes.1579    Mrs. Biggs. Thank you, Chairman Weber, and thank you to our1580witnesses for being here today.1581    The growing energy demand of AI infrastructure requires 24/15827, 365-day baseload power generation that never goes dark. So1583do our American businesses and consumers. Nuclear energy is the1584solution. It provides clean, efficient, and resilient power1585that keeps the lights on and the rates low.1586    South Carolina is already a leader in nuclear. Over 501587percent of power generated in South Carolina comes from a1588nuclear plant. I'm blessed to live just a few miles from one of1589the largest nuclear power stations in the United States. The1590Oconee Nuclear Station has provided reliable power to the 3d1591District for over 50 years, and it was recently renewed for1592another 20 years.1593    South Carolina is an exciting place to be. We have a record1594economic and population growth. However, we are also reaching a1595point of energy criticality. South Carolina needs the kind of1596reliable and resilient energy production that nuclear does1597provide. The newly formed Palmetto Nuclear Coalition was1598launched with the goal of bringing the nuclear renaissance to1599South Carolina, whether that is in the form of traditional1600reactors or small modular reactors. So my question is to Mr.1601Schweiger.1602    The energy demand from data centers and manufacturing is1603only increasing, while American baseload has stagnated over the1604last 20 years. What Federal policies could help scale nuclear1605capacity fast enough to meet and exceed the growing energy1606demand?1607    Mr. Schweiger. OK, thank you for the question. That's a bit1608of a toughie.1609    I think that--so I grew up in Washington State. There were1610five nuclear plants that were under construction. Only two--1611actually, one--got built. Four were canceled. And so, when you1612look at what happened, it was--at the State level they didn't1613have the funding duration to support all five reactors.1614    So I think what's crucial is that the U.S. industry moves1615at pace. We've heard other panel members here talk about--or1616witnesses here talk about how quickly we can get a plant built.1617So the industry has to move faster in America. And then there1618has to be the money to back the initiatives.1619    Mrs. Biggs. Thank you.1620    For Ms. Barron, how could the licensing process be1621simplified to allow for quicker project initiation to power1622generation?1623    Ms. Barron. Thank you, Representative Biggs, for that1624question. And, thankfully, that has been a subject of much1625attention of late, with the President's executive orders1626focusing on streamlining the relicensing and licensing1627timeline.1628    There is no question that we need a very competent and very1629responsible Federal regulator to be overseeing the industry,1630but we also need to move as quick as we possibly can, making1631sure that we meet all, you know, regulatory and safety1632requirements and we reduce unnecessary regulation.1633    Like, for example, for us to get an early site permit1634renewed at our Clinton site costs about $35 million, take a1635couple of years to evaluate whether that site is suitable for1636nuclear power when it already has a reactor on the site. Like,1637these are the kinds of things we're trying to point out that we1638could reduce the unnecessary work, focus on the necessary work,1639and do our work as quickly as possible. We can achieve the goal1640of getting the reactors online faster.1641    Mrs. Biggs. Great. So we're on a roll. I'm just going to1642continue with you. What is the most effective role for the1643Federal Government to aid in the production and scaling of1644small modular reactors?1645    And how might they be useful for building out AI1646infrastructure and keeping our rates low?1647    Ms. Barron. Well, in my testimony I mentioned some critical1648policies including 45Y and 48E tax credits to support1649investment in--in new--new reactors. I mentioned the Loan1650Programs Office, which is another very important tool. And then1651there are some grant programs that are underway, both for SMRs1652and, ideally, for other large-scale, new reactors, as well.1653Depending on the use case you might prefer to have a larger1654reactor, as opposed to a smaller reactor. And of course, the1655larger reactor, the AP1000, has already been successfully1656deployed in Georgia, and so it doesn't have to go through that1657same sort of licensing as--as the newer reactor designs.1658    But all of these programs obviously have to work together.1659    Mrs. Biggs. Thank you so much for your insight.1660    And with that I yield back.1661    Chairman Weber. I thank the gentlelady. The Chair1662recognizes the gentlelady from Oregon for at least 5 minutes.1663    Ms. Salinas. Thank you, Mr. Chair, and thank you to our1664Ranking Member and our witnesses for being here today.1665    Whether we're using nuclear on different power sources, I1666think we need to make sure that we also maintain focus on1667efficiency, and this kind of goes to the question that our1668Ranking Member of the Full Committee was having.1669    If we can limit data center power needs in the first place,1670that will make it easier to lead the world in AI while keeping1671costs under control for our electric grid. For example, in1672Oregon, the Corvallis Microfluidics Tech Hub is a consortium1673tackling R&D to make chips more efficient and easier to cool.1674    Dr. Renshaw--and again, you had a little bit of this1675conversation--you alluded to this in your testimony. Can you1676elaborate on the R&D being done in this space, and how energy1677demand projections change depending on what we are able to1678accomplish in that efficiency space?1679    Dr. Renshaw. Yes, so great question. I would say there are1680many opportunities in this space.1681    So you had mentioned advances in chip designs, which we've1682seen significant improvements in efficiency, thousands of times1683of improvements of the number of tokens or word portions that1684we can generate per unit of energy. Additionally, what we've1685seen is that as new models, new model architectures for AI are1686developed, they're often more efficient and more performant.1687    One of the things that we're doing at EPRI right now is1688partnering with others in industry to look at how can we use1689domain-specific models, so models that are customized for a1690particular task to be able to not only get better performance,1691but use less energy in combination.1692    So we think that all of these together will help to at1693least blunt the growth of artificial intelligence energy needs.1694    Ms. Salinas. Thank you.1695    Ms. Barron--and again, teeing off of the conversation with1696Ms. Biggs--while meeting the significant energy demands of AI1697infrastructure is important, I too worry that ratepayers could1698end up with higher energy prices along the way. Past examples1699of nuclear energy construction projects have either failed or1700were completely after long--completed after long delays and1701cost overruns. And while I support the need for clean energy to1702be added to the grid, I want to make sure that ratepayers are1703not flipping the bill and subsidizing the costs as risks of1704powering these data centers continue.1705    Can you describe how Constellation is protecting ratepayers1706from unnecessary costs?1707    Ms. Barron. Thank you for the question. And, you know, I--1708as we talked about earlier in response to Ranking Member Ross,1709there's no question when you're deploying a first-of-a-kind1710technology that there are going to be challenges. And we've1711seen that with, for example, the offshore wind development over1712here on the East Coast. But--but once the resources reach nth-1713of-a-kind, and you can get some more predictability, then it's1714easier to manage the cost, of course.1715    I mean, I think it's true that, if you look at the Georgia1716example, Vogtle Unit 4 was 30 percent cheaper than Vogtle Unit17173. And so when you can get an order book in place and you can1718get to that nth-of-a-kind spot, obviously this is all easier.1719But in the short term it is challenging.1720    In our case, given that we don't have captive ratepayers1721and we're in competitive markets, what will likely happen is1722that we will have some corporate off-taker, likely some type of1723data economy customer, who will say I want to work with you to1724develop that resource, and I will take the power when it's1725done. And so there will be no ratepayer impact of that1726technology investment if it happens. But there still is a lot1727of work to get to that point. And that's what we're working1728hard to do.1729    Ms. Salinas. All right. And just as a follow up, are you1730worried that a reduction in LPO's ability to provide loan1731guarantees for large energy infrastructure projects, whether as1732a result of staff reductions or funding recisions, will1733inadvertently lead to an increase in ratepayer costs?1734    Ms. Barron. I mean, there's no question that the LPO, when1735it provides loans, can help bring down the costs of these new1736technologies, and that has benefits across the country for all1737ratepayers when you have a technology that can get1738commercialized at a lower cost. So we have been pleased to see1739the focus on continuing the investment through the LPO, both in1740small modular reactors and larger reactors through that1741program.1742    Ms. Salinas. Thank you.1743    I yield back.1744    Chairman Weber. The gentlelady yields back. The Chair now1745recognizes the gentleman from Indiana for 5 minutes.1746    Mr. Baird. Thank you, Mr. Chairman and Ranking Member, and1747thank the witnesses for being here today.1748    Ms. Barron, I'm going to start with you. And Constellation1749was among the first major utilities to invest in grid scale1750small reactor technology. And with this backing from Rolls1751Royce, the SMR in 2020, this endorsement from a credible1752nuclear utility served to inspire investment in the space,1753unlocking capital flow into several other SMR designs.1754    And with the President last month setting a goal of1755expanding America's nuclear capacity and capability from 1001756gigawatts to 400 gigawatts by 2050, the moment again calls for1757major credible first movers to help America win the race and1758drive deployment commitments from small modular reactors.1759    So my question really comes--since we're in this race, what1760policy hurdles or other impediments that inhibit or prevent1761major nuclear operators such as yourself from transitioning1762investment in SMR designs to the actual deployment of these1763SMRs?1764    Ms. Barron. Thank you for the question, Representative. And1765you're right. We did make an early equity investment in the1766Rolls Royce SMR, and that might be confusing to some because,1767of course, we're here in the United States. And why did we do1768that? But we did that because the UK government made an1769investment in that technology and launched a process to seek1770input and bids to award contracts to a large number of SMRs for1771use by the UK citizens. And we saw that as a signal that the1772government was supporting that technology, and that there was a1773future for Rolls Royce and potentially other SMR developers in1774the UK. And we're pleased to see that that has borne out the1775case, that the UK has selected Rolls Royce to move forward.1776    So, you know, what's happened in the intervening years1777since we made that investment is that we've had tremendous1778support here in the U.S., and we have a lot of very promising1779designs that are underway. And so we're--we're hopeful that the1780U.S. will--will see that same level of investment both in the1781work that Oklo is doing with the military and with commercial1782customers across--across the country.1783    Mr. Baird. So I'm going to continue on somewhat in that1784vein. In April, Representative Harrigan, Tenney, and I1785introduced the Small Modular Reactor Commercialization Act, and1786that was aimed at securing United States' preeminent position1787to industrialize grid scale small modular reactor technology.1788The bill amends the outdated, arbitrary 300 megawatt threshold1789for SMRs, which has really disadvantaged the U.S., and1790established a working group to continuously recommend policy1791that protects American status as the most competitive Nation1792for reactor companies to base manufacturing beyond first-of-a-1793kind demonstrator volumes.1794    So from your perspective, would you elaborate on what it1795takes and on this bill in order to move and have the workforce1796that we need to commercialize these SMRs?1797    Ms. Barron. Did you want to cover that?1798    Mr. Schweiger. Who are you directing the question to?1799    Ms. Barron. I'm happy to answer, but----1800    Mr. Baird. Now that we've taken this time, all three of1801you. But I'm going to start with Ms. Barron.1802    Ms. Barron. I'm not--I'm not familiar with every provision1803of the bill, but I understand that it's designed to help the1804SMR industry. So I, you know--and sort of modernize the code in1805order to--to ensure that it can be commercialized. So that's an1806important step.1807    I think there's also steps that the industry needs to take1808on the workforce question to make sure that we're helping1809support certification programs, bringing students into--from1810high school, through the trades, through the 4-year schools,1811into industry to--to power it moving forward. And we are doing1812that at Constellation.1813    Mr. Baird. Mr. Schweiger?1814    Mr. Schweiger. So when I was young, which was a few years1815ago, at the plant I worked at what became crucial is that the1816plant design was basic enough where you didn't have to have a1817Ph.D. to run it. So I think one of the keys in getting these1818SMRs to market beyond first-of-a-kind is to make sure the1819designs are as simple as possible so they're cost effective,1820easier to build, easier to run.1821    Mr. Baird. Dr. Renshaw.1822    Dr. Renshaw. Yes, I would agree with what has been said so1823far.1824    If I can add one piece of additional knowledge, I would say1825that training is an important area, and the workforce that we1826would need in the future for supporting nuclear is1827significantly larger than what it is today. This might be an1828opportunity to utilize AI to accelerate training proficiency,1829to be able to help this--the new generation to understand the1830technologies in these areas faster, to be able to get up to1831speed, to be able to replace the current workforce--or not1832replace, but augment the current workforce.1833    Mr. Baird. I thank all of you and I yield back. My time is1834up.1835    Chairman Weber. The gentleman yields back. The Chair1836recognizes the gentlelady from California for at least 51837minutes.1838    Ms. Friedman. Thank you very much, Chair Weber and Ranking1839Member Ross, and for the witnesses coming here today.1840    My constituents in California are really struggling with1841surging electrical rates: 1 in 5 ratepayers are behind on their1842power bills as of last year, and rates are more than 80 percent1843higher than the national average. Our State is home to more1844than 270 data centers, with 70 in Los Angeles alone and two in1845downtown--and in downtown Los Angeles, real estate developers1846are racing to build even more data centers to keep up with1847demand.1848    As AI-driven data centers and that demand surges, it's1849really important that we make sure that our ratepayers can also1850pay for the cost of electricity. Given how data intensive or1851how energy intensive these AI centers are, and given that1852California is one of the homes of a lot of this technology, how1853do we make sure, No. 1--well, my question is, are we building1854new power capacity, really, to satisfy the demands of data1855centers more than of ordinary citizens and their homes and1856their businesses?1857    And how do we ensure that none of those costs are being1858passed along to ratepayers?1859    [Pause.]1860    Ms. Friedman. And I don't know if you can speak to that,1861because I think maybe you're more on the technical side, but1862yes, go ahead.1863    Ms. Barron. I'm happy to try. I'm not an expert on1864California, per se, but I do think--and there was a prior1865question about, you know, are we going to see this sustained1866level of--of demand growth as the AI industry evolves, and I1867think that is--that is a good question.1868    But I also think, if we are going to continue to electrify1869our economy, if we're going to continue to try to onshore more1870manufacturing, and if we are going to continue to lose coal1871plants that go off the system, we are going to need to find new1872technologies to bring onto the grid. And--and all of that is1873going to benefit California and the country if we can do that1874successfully.1875    But we should be choosing the lowest-cost solutions. We1876should not be choosing things that we prefer because we like1877those technologies. We should be choosing the things that can1878achieve the goal of clean and reliable electricity at the1879lowest cost. And, you know, I think that's what we're all up1880here trying to do, and so that's what I would say to that1881question.1882    Ms. Friedman. Right now we--you know, in California I don't1883think that new nuclear is even allowed by law at this point.1884    Ms. Barron. That's right.1885    Ms. Friedman. We have Diablo Canyon, which is an older1886plant, which is--whose life has been extended recently through1887legislation which is actually tremendously costly to continue1888to operate.1889    But in terms of other parts of the country where new1890nuclear is allowed, you know, we hear a lot about permitting1891and how difficult permitting it is. And I can only imagine for1892a nuclear plant how many layers of safety review you have to1893do, and different levels of permitting. I'm curious as to1894whether anyone has ever done an analysis as to how much of that1895permitting may be repetitive, if there is permitting1896requirements that are--that could be streamlined, that are more1897low-hanging fruit to get things moved along more quickly and in1898a more cost effective way without sacrificing safety, community1899input, siting issues, that kind of thing.1900    Are there--if we're interested in making the permitting a1901little bit--have more sense to it and be easier for those1902communities who want to add new advanced energy and nuclear,1903where would we look at that?1904    And have you seen these kinds of--I mean, are--when you're1905doing the permitting, are you pulling your hair out saying I've1906already done this analysis, why do I have to do it again?1907    Ms. Barron. Well, I mean, I gave the example of the early1908site permit that we have, where we have to evaluate the1909environmental and other seismic issues associated with the new1910reactor, even though we're looking at the exact same site where1911we have existing reactors. So those kinds of things clearly are1912in focus to make sure that those are not things that we're1913wasting time on when we should be spending time on--on trying1914to get these newer designs certificated.1915    But I also think it's one of the reasons why using an1916existing site makes some more sense. You already have a cooling1917lake there, you don't need to build a new one. You have rail,1918you have electrical infrastructure, and those can bring down1919the cost and shorten the time of bringing on new reactors.1920    Ms. Friedman. Does anyone else have any thoughts as to what1921you would look at for anyone here who's looking at any kind of1922permitting, streamlining or reform?1923    And it's OK if you don't know.1924    Mr. Schweiger. Well, I think--so I've worked in fusion and1925fission, and the U.S. has policies in place that are making1926fusion permitting go faster. And so, you know, the fresh look1927at how to go get a nuclear-related technology to market, you1928know, to power the grid, there's already an example where the1929U.S. is--is moving faster with permitting.1930    So--and in fission there is some momentum gaining here, but1931I think it's important to support that and give it impetus.1932    Dr. Renshaw. And if I could add just two thoughts on this,1933as well--I know we're close on time--the Nuclear Regulatory1934Commission has already performed internal reviews on how they1935can streamline their own processes and procedures. That's been1936ongoing for several years now.1937    One thing that we could also utilize--not to beat a dead1938horse--but AI. AI is very good at streamlining and distilling1939large volumes of information. You may have hundreds of pages in1940a report. The person who's reviewing that report may only need1941certain key pieces of information. So instead of reading the1942entire report, could you utilize AI to accelerate the reviews1943of--of the permitting processes and so forth to be able to1944streamline on both sides of the review and preparation?1945    Ms. Friedman. Thank you, I yield back.1946    Chairman Weber. The gentlelady yields back. The Chair now1947recognizes gentleman from Colorado, Mr. Hurd, for 5 minutes.1948    Mr. Hurd. Thank you very much, Mr. Chairman, for convening1949this meeting on a very important topic.1950    The Trump Administration has signaled that nuclear energy1951is a key part of achieving energy dominance. The President's1952executive orders on nuclear energy, I think, send a strong1953message that America must lead the world in nuclear energy. And1954his executive orders focus on the nuclear supply chain,1955existing reactors, new reactors, and much more. However, what1956will matter is how they are implemented. We have to make sure1957that they're implemented in a way that creates predictability1958for companies so that they can raise private capital, but also1959allows projects to be built faster.1960    Mr. Schweiger, Oklo's CEO (Chief Executive Officer) was at1961the executive order signing ceremony, if I'm correct. How do1962these EOs impact Oklo's plans to build your reactors?1963    Mr. Schweiger. Well, they'll certainly help. Again, going1964back to the business model where Oklo is--is raising their own1965capital, building their own plants, and then selling that1966power, the EOs don't have a direct impact on Oklo as a company.1967I think there are--to the nuclear industry in general, it's1968great to have some tailwinds moving it along.1969    Mr. Hurd. Good. I'm happy to hear that.1970    Ms. Barron, I think Constellation's CEO was also at that1971executive order signing ceremony. Similar question: how do1972these executive orders impact Constellation's plans related to1973maintaining your current nuclear fleet and also for building1974new reactors?1975    Ms. Barron. Thank you for the question. That is true, my1976boss was there.1977    And I think the--the focus on the timelines, both for1978license extensions, which we talked about, how important they1979are to allow the existing fleet to continue operating for 201980years, streamlining that process, and looking at the reactor1981oversight process and the reactor security rules which haven't1982been updated in a long period of time, those are some important1983pieces of this executive order, although there are many that1984relate to workforce and other things, as you mentioned.1985    We are not currently developing a new reactor at this1986point, but we're looking at what that would take. And so having1987this streamlining in the EOs is going to make a--is going to be1988a big help.1989    Mr. Hurd. What do you think is the most challenging part of1990the EO to implement? Does anything come top of mind, Ms. Barron1991or Mr. Schweiger or Mr. Renshaw?1992    Ms. Barron. Well, as I mentioned in my testimony, there are1993some ambitious goals. I mean, 10 new large reactors under1994construction by 2030. It's ambitious, it's appropriately1995ambitious, but we have a lot of work to do to make that happen.1996    Mr. Hurd. Mr. Schweiger, the most challenging part of these1997EOs to implement, any thoughts?1998    Mr. Schweiger. None at the moment.1999    Mr. Hurd. OK, Mr. Renshaw, how about you? Anything to2000contribute here with respect to the President's executive2001orders?2002    Or maybe what role should Congress play in making these2003executive orders successful and impactful?2004    Dr. Renshaw. I'll go back to EPRI generally doesn't comment2005on policy or directions that Congress should take, so I will2006refrain from answering.2007    Mr. Hurd. OK, fair enough.2008    I want to ask--this is a question for all of you, so let's2009see who wants to answer. Given the bipartisan support for2010expanding our nuclear energy generation fleet and the taxpayer2011money that's gone into developing and fostering this industry,2012what barriers is the industry facing, and how can Congress help2013remove them so we can start building more reactors and putting2014electrons on the grid?2015    Mr. Schweiger. Thank you for the question.2016    I think the one barrier I see right now is just fuel2017supply, you know, having enough fuel. There's a lot of plants2018that need HALEU that are in the works. And so whatever can be2019done to get that fuel released for use, and then funding the2020research to be able to undergird what these--the new generation2021plants are trying to bring to market.2022    Mr. Hurd. Can I ask--do you see a future--do any of you see2023a future where spent fuel could be reused or reprocessed2024economically and securely in the United States, or is that not2025something that you see on the horizon?2026    Mr. Schweiger. So Oklo is doing that. We're working with2027Idaho National Lab right now to recycle the EBR-II fuel. And2028then one of the next objectives at Oklo is to go to the2029commercial spent fuel and recycle that. So it's very much in2030our program. We're seeing support from Congress, the U.S.2031Government, and we think it's a wonderful thing because there's2032a lot of spent fuel that could be used before it's buried.2033    Mr. Hurd. Ms. Barron, do you see any--with advanced2034reactors coming online, do you see any role that spent fuel2035inventories might play in fueling those systems?2036    Ms. Barron. I'm learning along with you about the Oklo2037design and the promise of that potential. I know that that is2038occurring overseas, like, for example, in France, but we2039haven't done that here in the U.S. And it would be a great2040development if that could be a source of power for a new2041reactor design, yes.2042    Mr. Hurd. Something to consider, indeed.2043    Mr. Chairman, I see my time has expired. I yield back.2044    Chairman Weber. The gentleman yields back. The Chairman2045recognizes the gentlelady from North Carolina for at least 52046minutes.2047    Mrs. Foushee. Thank you to our witnesses for being here2048today and for--to our Chair and Ranking Member for holding this2049hearing.2050    Dr. Renshaw, recent research published by Caltech and UC2051(University of California) Riverside looks at the fact that2052AI's environmental footprint can be disproportionately higher2053in certain regions, and raises questions about how we should2054fairly balance AI's rapid expansion with its regional2055environmental impact. Can you discuss how AI's environmental2056impacts can vary by region, and how--and should Congress2057consider addressing potential environmental inequities posed by2058AI's infrastructure?2059    Dr. Renshaw. So that is a very deep question, so we'll only2060scratch the surface on that today. But certainly there are2061disproportionate impacts, especially on under-privileged2062communities, because they're often sited in less desirable2063areas that may be close to power generation, facilities that2064may be more higher polluting or otherwise. And so we often look2065at the benefits of AI, but this is looking at kind of what is2066the flip side. So looking at the water use, the energy use are2067all things that we need to take into account.2068    I would say, on the positive side, as we continue forward2069on the path of research and development of these technologies2070to utilize advanced chips, more performant model architectures,2071domain-specific models, and advanced computing modalities such2072as neuromorphic and quantum computing, I think we can blunt2073some of that energy demand from data centers.2074    And we also have to remember on the environmental side that2075while the largest impacts are in the immediate area, certainly2076pollution can expand to greater regional areas, and pollution2077can, of course, cross cities and State lines. So we have to be2078cognizant that it affects all of us.2079    Mrs. Foushee. Thank you for sharing that insight.2080    Ms. Barron, I'm proud that our AI task force report from2081last Congress dedicates an entire chapter to discussing AI's2082energy and environmental impacts. As I've mentioned before, I'm2083concerned about how the environmental demands like water and2084local land resources needed to support AI's rapid arrival are2085affecting prior and recent commitments of our Nation's leading2086technology firms to become net zero emissions by the year 2030.2087    As we move into the second half of this decade and approach20882030, how do you assess the current capabilities of our2089Nation's clean energy infrastructure to support AI's continuing2090expansion today?2091    And what should technology firms and Congress be doing now2092to ensure that AI technology development into the future is2093sustainable? I think you touched on a bit of that earlier.2094    Ms. Barron. I did, but I appreciate that question. And in2095response as well to your last question, I guess I would I would2096say two things.2097    One is, you know, we've talked about this notion of2098collocation. You know, our plants tend to be very remote from2099population centers, and they have enormous land buffers. In2100most cases we have thousands of acres of land around the plant,2101and people don't even know that the plants are there. But that2102does make it an ideal location for locating a data center,2103which likely, you know, may cause some of the same concerns in2104that people don't usually want to look at that all day long, or2105hear it in some cases.2106    But to your exact question, addressing that land use issue2107by collocating with an existing plant and addressing the water2108issue, as well--as you may know, we have to create cooling2109lakes or cooling ponds to--to have water to cool our reactor.2110And in some cases it may be the case that the data center can2111use some of the water that we have already created for use by2112the plant. Sort of discharge water that we use for cooling they2113can also use for cooling. So there will be no impact on the2114local community for the water side, as well.2115    And then just to the last question on the potential for2116using AI to help with clean energy infrastructure development,2117one example I can give you is in addition to our nuclear2118reactors we own the largest hydro dam east of the Mississippi2119at the Conowingo up on the Susquehanna. And we also own a2120pumped storage facility that's right nearby. And these two2121facilities both use the water from the river to create2122electricity. But using AI, we've been able to optimize the use2123of the water in a way that we haven't been before that allows2124us to make 250,000 megawatt hours from that plant, which is2125enough to power 25,000 homes. That's more renewable energy we2126didn't have before we used this technology.2127    So a small example, but we're hoping those kinds of2128things--we can come back to you with--with--with even more2129impactful examples in the future.2130    Mrs. Foushee. Thank you, and my final question to Dr.2131Renshaw.2132    As a fellow North Carolinian, you know my district is a2133leader in researching and developing emerging technologies like2134quantum and AI. What opportunities are there in the upcoming2135reauthorization of both the National Quantum Initiative Act and2136the National AI Act of 2020 to advance U.S. leadership at the2137intersection of technologies like quantum and AI?2138    Dr. Renshaw. Yes. So I would say that AI and quantum are2139two of the most exciting technologies today. They are both2140exponentially growing technologies. AI is one that everyone in2141this room and pretty much around the world is aware of. Quantum2142is still a little bit under the radar, but it is growing very2143rapidly and very quickly. Certainly in the Research Triangle2144area of North Carolina, here in the D.C. area there are top2145notch universities, as well, as well as many other areas around2146the country and around the world.2147    So I would expect that as we continue to advance research2148and development in the areas and then move to practical2149applications, we will start to see not only the benefits of AI2150and the benefits of quantum, but the benefits of merging these2151two technologies together for both more performant systems,2152tools, and models, as well as more energy efficient systems,2153tools, and models.2154    Mrs. Foushee. Thank you.2155    That's my time. Thanks, Mr. Chair.2156    Chairman Weber. You bet. The Chair now recognizes the2157gentlelady from Maryland for at least 5 minutes.2158    Mrs. McClain Delaney. Thank you to the Chair and the2159Ranking Member, and thanks to our witnesses. And I know it's2160been a--already a very weighty morning, but incredible2161testimony so far.2162    So many of you highlighted earlier in your testimony and2163your answers about how the U.S. is obviously barreling toward2164this energy--I hate to use the word ``crisis,'' but it's really2165something that we've got to address--and that our consumption2166is increasing, driven by AI and our data center expansion. And2167without additional power generation from a variety of sources,2168including nuclear, many of us really are concerned about how2169costs will increase substantially for consumers and companies.2170    As we look to the 2030s, one way to drive down costs and2171increase nuclear power generation is the next generation of2172nuclear reactors deployed at scale. And as was noted by our2173Chair, one of the companies highlighted in the preparatory2174materials is my own 6th District's X-energy headquarters. And2175with the Department of Energy's Advanced Reactor Demonstrations2176Project and program, X-energy has partnered with Dow Chemical2177to provide a first-of-its-kind deployment of an advanced2178reactor that will generate both electricity and steam for Dow's2179chemical production behind the meter.2180    This initial demonstration was deemed critical to attract2181the next customers, and it's an approach that is working as2182Amazon has stepped up to partner with X-energy based on the2183ARDP to invest another $334 million in a second plant and2184target 5 gigawatts of new power. So a couple of things.2185    Dr. Renshaw, it appears that the new business models that2186are helping to deploy these small modular reactors at a rate2187that will bring down the initial capital investment needed for2188new generation of energy. How does the collaboration between2189high-need energy consumers and nuclear energy producers like X-2190energy's partnerships with Amazon and Dow Chemical meet the2191needs of AI companies while keeping energy prices down for U.S.2192consumers?2193    And we talked about that a little bit earlier, but kind of2194delving a little bit into it because I was very impressed with2195X-energy.2196    Dr. Renshaw. Yes, if we can maybe peel the onion back one2197more layer to get a little bit deeper, I would say one of the2198key things that's holding us back today is really the financing2199aspect. The cost of financing for all of----2200    Mrs. McClain Delaney. I completely agree with that.2201    Dr. Renshaw [continuing]. These projects is really the2202largest driver.2203    So as we can move from pilot phases to demonstration and2204then production--I heard nth-of-a-kind earlier--we can reduce2205the cost, build that trust and credibility as we take the2206fundamental R&D to applied R&D into production, and that will2207help everyone--not only X-energy, but other reactor vendors--to2208be able to build advanced nuclear, as well as other2209technologies that are clean, safe, affordable, and reliable,2210and put those on the grid for the benefit of all.2211    Mrs. McClain Delaney. So building on that, would you say2212that would be something that would be through a public-private2213partnership? Do you see it in terms of private equity markets2214getting into it? Or do you think it's something that we really2215need to step up, as--you know, Congress, and looking at funding2216it, you know, more from a government level as we look at it2217ahead? Because I think that's incredibly important.2218    Dr. Renshaw. Yes. In this case I would say all of the above2219are important and helpful. I would say the people who are2220building these systems, they--they would need support and2221collaboration partnerships to be able to make them happen. I2222think my--my co-witnesses can testify to that. And in fact, I2223might defer to them if you want more details on what would be2224important----2225    Mrs. McClain Delaney. No, I would love to hear from them2226and--briefly. Then I want to get on to research for a second,2227as well.2228    Ms. Barron. Yes, I think that's exactly right. I mean, I2229think the stability of the government policy is important, and2230knowing what tools are available is sort of what everyone is2231looking at at this moment in time.2232    But--but once you have that certainty, then you are going2233to need to go to capital markets, you're going to have to bring2234other investors along, you're going to have to understand what2235their cost of capital is and how you're going to share the2236risk. And then you put it together and then you go. I mean,2237that is--that is there's a lot of folks working on how to make2238that happen right now.2239    But, you know, the offtaker matters, the government--the2240government policy matters. But we are going to need to look to2241third party capital, as well.2242    Mr. Schweiger. OK. Finally, Oklo's perspective, the Loan2243Programs Office, this is an important financial tool for Oklo,2244we want to see it presented. Secretary Wright supports this2245program, as well.2246    Mrs. McClain Delaney. So thank you all. I'm just going to2247submit for the record a question that, you know, the budget2248request cut DOE--the President's budget cut DOE's ARDP program2249by 51 percent, totaling $161 million. And I'm really concerned2250about the implications of reduced research funding and nuclear2251power generation, that fundamental research. So I'll submit for2252the record.2253    But thank you all, and very informative.2254    I yield back.2255    Chairman Weber. The gentlelady yields back. The Chair now2256recognizes the gentleman from Illinois for at least 5 minutes.2257    Mr. Foster. At least? I'd like to understand the legal2258reading of that, and what--the penalties involved for--anyway,2259one of----2260    Chairman Weber. Ms. Barron is an attorney, if you're2261interested.2262    Mr. Foster. OK. Well, since I haven't yet been assigned a2263Subcommittee here, I think you have considerable leverage in2264that negotiation over me.2265    One of the areas where AI is going to have a real impact, I2266hope, is on the compliance and engineering costs for--for2267nuclear that--you know, I was talking to a guy who actually2268runs a company that builds a lot of these data centers. And2269when you deliver the civil construction for a data center, it's2270accompanied by just a telephone book full of documentation on2271how your, you know, tornado resilience is, and you name it.2272    And so--and he's in the process of replacing a very large2273group of engineers and--and so on that produces that phone book2274with AI, because once you've done that for one data center,2275trained it on things, you can very rapidly make the minor2276modifications in this.2277    And so in the case of, you know, nuclear, if you--if you2278train your AI on every site evacuation plan that's ever been2279written, and say I need a site evacuation plan for one more2280place, you can imagine that that's an instance where the2281engineering costs associated with a new nuclear emplacement may2282go down.2283    Also, the regulatory delays if that same level of AI is2284used by the government to evaluate this telephone book that2285you've just submitted to them, then you could imagine a very2286rapid turnaround in that. I was wondering. Are you starting to2287use, you know, AI to generate any of the paperwork or--the2288electronic paperwork so far for your regulatory things?2289    Mr. Schweiger. So we have an AI protocol that's in writing2290for Oklo, and we're starting to use AI in expanding capacity.2291When you start looking at design of a first-of-a-kind plant, I2292think you can use AI to a limited capacity. But there is a lot2293of details that go into how, for example, a heat exchanger is2294designed.2295    Mr. Foster. OK, but I was referring to the site-specific2296details.2297    Mr. Schweiger. OK.2298    Mr. Foster. That when you have--OK, I have one of these----2299    Mr. Schweiger. Oh, yes.2300    Mr. Foster. You know, because one of the messes that you2301have in, like, all the Constellation plants is that they're all2302somewhat different and, you know, the cooling scenario is2303different and the ponds are different. But that is something2304where, once you had the fundamental engineering understood by2305the AI, you could rapidly make a new site-specific plan. And it2306would, I think, change a lot of the economics. And if the2307government would go--move along with you, you know, the2308approval of that could be a lot faster.2309    And so anyway, I just urge you to keep your eye on that,2310because I think that--you know, the--it also means that the2311workforce planning will be a lot--you know, the--the group that2312is anticipated by this guy that makes the AI data centers was2313about 5 percent of this current group size--could produce all2314of that paperwork using AI trained on--on projects that have2315already been completed.2316    Let's see, one--all right. Another thing, you know, I2317worked at Fermilab for many years. We're a very good customer.2318And one of the deals that we had was that we would get a call2319from the, you know, control room, and said, hey, it's--you2320know, it's a hot afternoon. We're getting in kind of trouble on2321our capacity. Can you guys, like, do some preventative2322maintenance instead of drawing power? And that got us a much2323better rate, power rate. Are you seeing that same sort of thing2324out of the data centers?2325    Because the data centers, in principle, you know, they're2326doing a mixture of 2 week-long projects to do--to train models,2327and then rapid response to people that type in queries. The--2328the queries could be routed to any data center, you know, in2329the continental United States so that that load can be moved2330around. And you can certainly just say, OK, it's going to take23312 weeks and 2 days because we had an ice storm in Texas. And is2332that sort of--are you getting that sort of negotiation out of2333the data centers at this point?2334    Yes.2335    Dr. Renshaw. Yes, so I'll take maybe the first part and2336then defer to Ms. Barron for the second part.2337    So we are looking, through our DC Flex Initiative, of data2338center flexibility, how we could do exactly what you're talking2339about, utilizing the ability to delay AI training or move it to2340different sources or different locations. In some cases you're2341able to do that. In other cases you're not. So we want to be2342realistic about what can and can't be done.2343    On the flip side, there's also the opportunity, as2344discussed earlier, to be able to use backup generation sources.2345Now, currently many of those are diesel and can only be2346operated so many hours per year, which could still be a good2347resource, even for the top 40 highest peak load hours during2348the year. If cleaner energy sources were used such as hydro-2349treated vegetable oil, hydrogen sources, or other, then you2350could potentially use that backup generator as a grid resource2351more frequently.2352    So maybe I'll defer to Ms. Barron to talk more about that.2353    Ms. Barron. Well, I don't have anything to add. I think2354EPRI has shown tremendous leadership in helping the industry2355get to exactly your question, where is the untapped flexibility2356that we can use so that we can understand what they can do, and2357then we can design market rules that will provide the right2358incentives so that they do dial back at times when the system2359needs the power.2360    Mr. Foster. Yes. Well, the other thing I worry about is if2361you just look at it from a CapEx point of view of the data2362center operators, you know, you're asking them--they've put all2363this money into their GPUs, and you're asking them to let that2364CapEx sit idle for some period of time. And so there's a2365calculation that, in principle, you could do right now is--you2366know, at what point they say I don't care, you know, I don't2367care if I get a lower rate. It's--electricity isn't that big a2368deal for me. And that--so that part of the calculation I think2369we can understand now as to whether this is going to be a2370winning game for leveling the load on this.2371    Anyway, this is--I'm glad to see you're thinking about it,2372because this is, you know, sort of the first--the challenge2373that nuclear has, frankly, is $0.10 a watt Chinese--that's the2374spot price for Chinese solar panels. I recall, you know, 152375years ago, when we were arguing about cap and trade, we had2376these U.S. solar startups optimistically projecting a dollar a2377watt, which I was skeptical of. And now it's $0.10 a watt. And,2378you know, Goldman Sachs has done a big analysis of these and2379they find that at current prices the winner by far is a2380collocated solar field and battery and a--and a data center.2381    Chairman Weber. The gentleman's 15-year memory time has2382expired.2383    [Laughter.]2384    Chairman Weber. I now recognize the gentleman from Virginia2385for 5 minutes.2386    Mr. Beyer. Mr. Chairman, Mr. Chairman Weber, thank you so2387much for allowing me to waive on. It's--it's fun to be back2388here.2389    And to the witnesses, thank you very much for sitting in2390with us. I apologize for beating a dead horse, but I'm just2391extraordinarily concerned about the future of our country with2392the retraction in our investment in science and technology.2393    And Mr. Schweiger, I read you're head of engineering at2394Commonwealth Fusion Systems (CFS). We're very excited about2395everything that CFS is doing. I've been to visit a couple of2396times. We have our congressional Fusion Caucus. I'm a co-chair2397with another Democrat and two Republicans, and every--virtually2398everyone on this Committee is part of the Fusion Caucus.2399    So we were--and we're thrilled that Commonwealth Fusion has2400cut the deal with Dominion Power to build the first arc power2401plant in the history of humankind here in Virginia, but we're2402also concerned that China last year allocated $1.5 billion to2403fusion. We were at $790, the new budget is down 6 percent to2404$744 million. It's a 6-percent cut. In the CHIPS and Science2405Act we authorized $1.04 billion, which is still way short of2406where we need to be.2407    We met with Secretary Wright, our caucus co-chairs. The2408Secretary was very supportive, it was a great meeting. But we2409also know he doesn't control the budget. And it's not just the2410fusion budget. It's also National Science Foundation has been2411cut 55 percent, for example. What's happening at NIH (National2412Institutes of Health) is very sad, in Congresswoman Delaney's2413district, the--the so-called Bethesda declaration.2414    Mr. Schweiger, how do we ensure that fundamental enabling2415science that will propel energy technologies is actually2416prioritized, even in the current constrained funding2417environment?2418    Well, what do we do to make sure that we're putting our2419money where our dreams are?2420    Mr. Schweiger. OK, thank you for the question. It's a great2421one.2422    So I had the--I'll call it privilege--to go to China for 82423years off and on, and learned--learned a lot about how the2424Chinese think and how they run their programs. And I'd say--2425this is my view--but America is--is recognized as the premier2426country for ingenuity, thinking of new things and ways to do2427it. And so when you think about how we're going to implement--2428whether it's fission or fusion, my high bias is that Americans2429are the ones that are going to do it because we're ingenious.2430    And so to the question, the--I think fusion has a ways to2431go, right? There has not been a commercial fusion plant put2432together. It's a good thing, we're supporting that. The--the2433technology I'm representing is ready now. And so when you look2434at implementing technologies that will help, whether it's AI or2435just the power grid, I think America needs to step up. As my2436fellow witness said, you know, let's use what we already have.2437I think that's super important. And then let's get the new2438technology moving forward as fast as we can, by whatever means.2439As Dr. Renshaw said, everything above.2440    Mr. Beyer. Yes, yes. I think we all agree with as much as2441we can.2442    Dr. Renshaw, to you specifically, one of the cases that we2443made to Secretary Wright is, although we've seen an enormous2444amount of private-sector investment in fusion and in fission2445and others, which we--which we celebrate--I mean, TAE just2446raised $150 million over the weekend, which we're excited2447about--we still recognize that there are fundamental2448engineering and science problems that would serve the entire2449industry that can only likely be done by a Federal Government2450approach.2451    For example--and Mr. Schweiger, you know this from your CFS2452days--the high energy neutrons that are thrown off from a DT2453reaction, you know, dissolve metal. We need some way to--to2454figure out the--this--the engineering and the science that will2455allow us to deal with the largest single downturn.2456    Isn't that, Dr. Renshaw, the most appropriate place for2457Federal Government investment?2458    Dr. Renshaw. So I'll say that EPRI doesn't comment on what2459the policy should be, but I can comment on where is the2460technology going, and how can we get to an end state that is2461beneficial to all.2462    So we're looking at how can we bridge the gap between where2463we are now and where we need to be to be able to develop large-2464scale fusion plants. I would say that, similar to the other2465technologies that we mentioned in energy, there is no perfect2466energy technology, fusion included. But there are some2467significant benefits to deploying fusion.2468    So in--the ways that we can get there are similar to how2469research and investment has been done in the past, where the2470government is able to fund research and development programs to2471be able to foster that innovative culture that Mr. Schweiger2472had mentioned earlier. Some of the things that we're doing at2473EPRI are looking at how can we advance the state-of-the-art in2474fusion materials, as well as fusion plasma control. We actually2475recently completed a challenge for this, and we brought in2476organizations from around the world to compete on the best2477concepts for developing and deploying fusion technologies.2478    Mr. Beyer. By the way--and Mr. Chairman, just--one of the2479reasons we're so excited about milestones is because it2480emphasizes competition. You know, we're rewarding the2481entrepreneurs and the scientists and the engineers who can2482figure the problems out. Mr. Chairman, I yield back. Thank you.2483    Chairman Weber. The gentleman yields back. I thank the2484witnesses for your very valuable testimony and the Members for2485their questions.2486    The record will remain open for 10 days for additional2487comments and written questions from Members. The hearing is2488adjourned.2489    [Whereupon, at 11:58 a.m., the Committee was adjourned.]24902491                                Appendix24922493                              ----------24942495                   Answers to Post-Hearing Questions2496[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]24972498                                 [all]