Posted on September 18, 2024 Posted By: Danielle Forand Categories: Nimitz Tech

⚡NIMITZ TECH NEWS FLASH⚡

“Examining the Department of Energy’s Role in Advanced Computing Research”

Senate Committee on Energy and Natural Resources Hearing

September 12, 2024 (recording linked here)

HEARING INFORMATION

Witnesses and Written Testimony (linked):

Ms. Helena Fu: Director, US Department of Energy Office of Critical and Emerging Technologies.
Dr. Shaun Gleason: Director of Science-Security Initiative Integration,Oak Ridge National Laboratory Office of the Laboratory Director
Dr. Divyansh Kaushik: Senior Fellow,American Policy Ventures

Key words:

AI, national labs, china, PRC, quantum, Oak Ridge, research, NSF, supercomputer, energy.

IN THEIR WORDS

"American businesses doing business in the PRC... are inherently supporting their economy, inherently supporting an authoritarian regime... We cannot live in a dreamland of the China we want to operate with; we must recognize the China that we are operating with."

-Dr. Divyansh Kaushik

"We need to continue investing and leading these emerging technologies in a way that leverages DOE’s assets... We must secure our research from foreign espionage while continuing to leverage the best scientific minds in the world."

- Chairman Joe Manchin

"Making sure that data is AI-ready is a huge undertaking... You could have a lot of data, but it's not valuable until it's made AI-ready... data is the fuel for the AI engine."

- Dr. Shaun Gleason

OPENING STATEMENTS FROM THE FULL COMMITTEE

Chairman Manchin highlighted the importance of the hearing, which focused on the current state and future of advanced computing technologies, including artificial intelligence, quantum computing, and cybersecurity. The Chairman emphasized that while the U.S. leads in some areas, China is advancing rapidly in AI and other technologies, posing a risk to U.S. national security and economic prosperity. He discussed the bipartisan legislation introduced to support AI research at DOE labs, stressing the need for government involvement in critical technology development to maintain America’s leadership position.
Ranking Member Barrasso emphasized the critical role of advanced computing in maintaining America’s economic growth and national security. He noted that China is closely monitoring U.S. research efforts, specifically targeting technologies developed at national labs. The Ranking Member expressed concerns about the presence of Chinese nationals in U.S. labs, highlighting the risks of espionage and technology theft. He criticized the Department of Energy for not taking the China threat seriously enough and urged the department to protect American research from adversaries, warning that Congress may need to intervene if current measures remain inadequate.

WITNESS HIGHLIGHTS

Ms. Helena Fu, Director of the DOE Office of Critical and Emerging Technologies, highlighted DOE’s leadership in advanced computing, artificial intelligence, and quantum information science, emphasizing the department’s historical role in scientific advancement since the Manhattan Project. Ms. Fu described the DOE’s efforts in developing AI-ready data, advancing compute platforms, and creating quantum computing facilities, which are critical to addressing national security and energy challenges. She also noted the establishment of the Energy Threat Analysis Center (ETAC) to strengthen cybersecurity defenses for the energy sector and expressed her commitment to continuing the DOE’s mission in partnership with industry and international allies.
Dr. Shaun Gleason, Director of the Science Security Initiative at Oak Ridge National Laboratory, discussed the grand challenges in artificial intelligence, quantum science, and cybersecurity that the U.S. must overcome to maintain its technological leadership. He emphasized the need for energy-efficient AI systems and reliable quantum devices and pointed out the critical intersections of AI, cybersecurity, and quantum technologies. Dr. Gleason highlighted the importance of public-private partnerships and national labs’ experimental facilities in driving research and innovation, citing specific collaborations such as quantum communication experiments and AI model training with industry leaders. He concluded by stressing the need for continued investment and collaboration across sectors to address these technological challenges.
Dr. Divyansh Kaushik, Senior Fellow,American Policy Ventures, focused on the strategic competition between the U.S. and China in advanced computing and AI, noting China’s rapid R&D investment growth and espionage efforts. He outlined the significant role of DOE national laboratories in maintaining U.S. leadership and emphasized the need to secure these facilities from foreign exploitation. Dr. Kaushik warned that China’s strategy involves both legitimate R&D investments and the illicit acquisition of intellectual property, presenting a coordinated challenge to U.S. innovation. He stressed the importance of research security, advocating for policies that protect national interests without stifling innovation or targeting individuals based on ethnicity, and underscored the need for vigilant oversight to maintain America’s technological edge.

SUMMARY OF Q and A

Chairman Manchin asked the panelists to compare the roles of the Department of Energy (DOE) and the National Science Foundation (NSF), particularly in the context of advanced computing and AI. The Chairman expressed concerns about potential overlaps and inefficiencies between the two agencies, questioning if NSF might be duplicating DOE’s efforts, particularly regarding investments from the CHIPS Act. Ms. Helena Fu responded that DOE and NSF have complementary roles, with NSF focused on foundational research, workforce development, and grants to universities, while DOE focuses on large-scale, mission-driven R&D. She emphasized that DOE’s supercomputers have been made available to the broader research community in collaboration with NSF initiatives. Chairman Manchin further questioned if NSF was encroaching on DOE’s domain, potentially limiting DOE’s capabilities. Ms. Fu assured him that both agencies work in complementary ways, with DOE’s focus on specific challenges and NSF supporting broader foundational research. Dr. Gleason added that NSF provides essential access to computational resources and tools for students and universities, while DOE offers world-class scientific capabilities and expertise to help researchers achieve their objectives. Chairman Manchin expressed concern about the power supply needed to support AI and data centers, particularly with the significant retirement of dispatchable power sources like coal. He pointed out that while China is rapidly increasing its power capacity, the U.S. is facing challenges in meeting growing energy demands. Ms. Fu acknowledged this issue and mentioned DOE’s focus on grid resilience and energy efficiency initiatives, including studies on energy use in data centers. Dr. Kaushik highlighted the exponential growth in computational needs for AI systems, noting the increasing difficulty of connecting new data centers to the grid due to power supply limitations, which could impact the U.S.’s ability to remain competitive.
Ranking Member Barrasso asked Dr. Kaushik to explain the unique threats that China poses to government-funded research and development of critical and emerging technologies. Dr. Kaushik responded that China’s civil-military fusion policy, which mandates that all research, even at universities, ultimately serves military objectives, poses significant threats. He highlighted China’s legal requirements compelling researchers to share information with the government, which complicates their ability to act in good faith. He noted that China often targets cutting-edge technology for espionage when it cannot independently replicate it, exploiting the open research environment in the U.S.The Ranking Member referenced a report by the Hoover Institute on China’s “Seven Sons of National Defense,” a group of universities linked to military research, and asked why the DOE continues to collaborate with these institutions. Ms. Fu clarified that the DOE does not have any ongoing bilateral cooperation with the “Seven Sons” universities and is fully aware of the risks associated with China’s policies. She emphasized that the DOE maintains stringent research security measures, including foreign national screenings and bans on foreign government-sponsored talent recruitment plans, to mitigate these risks. The Ranking Member asked Dr. Kaushik if DOE should sever all ties with researchers from the “Seven Sons” universities. Dr. Kaushik firmly agreed, stating that no researcher funded by U.S. taxpayer dollars should collaborate with these institutions. The Ranking Member asked Dr. Kaushik what additional measures the DOE could implement to better protect advanced computing research. Dr. Kaushik commended DOE for being ahead of other agencies in implementing security measures but suggested that DOE and the broader research ecosystem need to increase transparency about the specific risks to national security. He recommended clearer boundaries between open and secure research environments and stronger protection of commercially relevant technologies developed in public-private partnerships. He emphasized the need for comprehensive guidelines to manage these partnerships effectively, particularly those involving significant commercial value. The Ranking Member concluded by discussing the U.S. energy challenges, referencing reports on the increasing power needs driven by advanced technologies like AI. He highlighted the significant contrast between the U.S., which has retired considerable coal power capacity, and China and India, which continue to expand their power production. He stressed the need to be clear-eyed about China’s ambitions to become a global superpower in military, economic, and technological domains.
Sen. Hickenlooper asked Dr. Kaushik about the electricity demands of AI and other emerging technologies and how much more energy these systems will require. Dr. Kaushik explained that predicting future demand is challenging due to the proprietary nature of companies’ plans for training new models. He noted that the International Energy Agency predicted that global electricity demand for data centers could double by 2026. He emphasized that even a smaller increase would require significant preparation, and private sector investments, such as a recent geothermal energy project in Colorado, indicate the growing need to address this challenge.Sen. Hickenlooper asked Ms. Fu about how the DOE collaborates with other agencies, such as NIST and NSF, to advance AI safety research. Ms. Fu described ongoing coordination efforts, including an MOU with the AI Safety Institute to leverage DOE’s expertise in AI safety and security. She highlighted that the DOE, particularly through its National Nuclear Security Administration, has been focused on evaluating AI models and developing safety measures. Additionally, DOE and NSF are working together to align workforce efforts and co-fund research on privacy-enhancing technologies, crucial for ensuring secure AI deployment. Sen. Hickenlooper inquired how the DOE plans to integrate different advanced computing technologies, including quantum and classical computers, to address complex scientific challenges. Dr. Gleason explained the importance of heterogeneous computing systems, which combine different technologies to tackle specific parts of a problem. He provided examples of past integrations, such as combining CPUs with GPUs, and highlighted the potential of quantum computers to handle optimization and molecular dynamics simulations. Dr. Gleason emphasized that DOE and national labs are focused on finding ways to connect and integrate diverse computing architectures to maximize their effectiveness in solving advanced scientific and technological problems.
Sen. Hawley asked Dr. Gleason to provide a brief overview of Oak Ridge National Laboratory’s contributions during the Manhattan Project. Dr. Gleason explained that Oak Ridge was established in the early 1940s to produce nuclear materials for the atomic bomb, including building the world’s first continuously operating nuclear reactor, the graphite reactor. He emphasized the lab’s enduring legacy in national security and material sciences, stemming from its roots in the Manhattan Project. Sen. Hawley highlighted the severe impact of improper nuclear waste disposal on communities near Oak Ridge and St. Louis, including radiation exposure. He praised the Senate’s bipartisan vote to compensate affected residents and urged the House of Representatives to pass the legislation, emphasizing the need to clean up these communities and provide compensation for their sacrifices during World War II and the Cold War. Sen. Hawley asked Dr. Kaushik to elaborate on the risks posed by American AI companies investing in and partnering with Chinese businesses. Dr. Kaushik stressed that American companies investing in China inherently support an authoritarian regime and potentially contribute to human rights abuses, including forced labor in Xinjiang. He emphasized the importance of recognizing the current realities of China’s government and the risks associated with supporting its economy, warning that U.S. businesses must be mindful of their impact. Sen. Hawley recounted a recent conversation with an Intel executive who downplayed security concerns related to the company’s investments in Chinese AI. He criticized the executive’s refusal to condemn the persecution of Uyghurs in Xinjiang, highlighting the moral and security hazards of American companies doing business in China. Sen. Hawley argued that companies benefiting from U.S. funding should uphold American values, such as opposing slave labor, and called for greater accountability among corporate leaders.
Sen. Cortez Masto highlighted the rapid increase in electricity demand due to AI and data centers, particularly in Nevada, a major hub for such facilities. She asked Ms. Fu for more information on how to address this emerging demand. Ms. Fu responded that the DOE is focused on this issue due to its leadership in advanced computing and energy. She noted that DOE is working to convene stakeholders, including those in high-load growth areas, to find solutions, and highlighted the department’s efforts to expedite and streamline the permitting process through AI innovations. Ms. Fu emphasized that DOE is addressing near-term, mid-term, and long-term needs to manage the growing energy demand from AI. Sen. Cortez Masto asked about DOE’s collaboration with other federal agencies, specifically the Department of Interior, which manages a significant amount of land in Nevada. Ms. Fu confirmed that DOE is working through the CEQ and partnering with 13 agencies on permitting processes, including exploring the potential of siting data centers on federal land. She stressed that DOE is considering all options to address the energy challenges posed by AI and data centers. Sen. Cortez Masto asked about the future workforce needed to lead in emerging technologies and what Congress should focus on to support this workforce. Dr. Kaushik responded that the competition for talent is fierce, especially given China’s large population and sophisticated talent recruitment programs. He emphasized the importance of the U.S. attracting and retaining global talent and highlighted the need for a comprehensive talent strategy. Dr. Kaushik pointed out that a significant number of top AI startups in the U.S. were founded by immigrants, underscoring the critical role of immigration in maintaining technological leadership. Sen. Cortez Masto agreed with Dr. Kaushik but stressed the need to balance securing sensitive technology with fostering innovation. Dr. Kaushik concurred, citing existing policies like the National Security Decision Directive 189 and NSPM 33 as frameworks that provide the necessary guardrails. He emphasized the need to classify sensitive research appropriately and manage access to ensure technological advantage, suggesting a careful approach that balances openness with security, tailored to the evolving nature of global competition.
Sen. King demonstrated AI’s impressive capabilities by reading a poem generated by ChatGPT about the U.S. Senate Committee on Energy and Natural Resources. He highlighted both the creativity and limitations of AI, noting a factual error about the committee’s members, which served as a reminder of the potential pitfalls of relying on AI-generated content without verification.Sen. King expressed concerns about potential duplication of efforts among U.S. government agencies working on quantum computing and advanced computing. He asked Ms. Fu for assurance that DOE is coordinating effectively with other agencies. Ms. Fu confirmed that DOE collaborates closely with the White House and other agencies, emphasizing that DOE focuses on large-scale scientific infrastructure, such as foundries and test beds, which are unique to the department’s capabilities.Sen. King asked how AI can assist in the efficient management of the power grid, emphasizing the need to distinguish between power generation and transmission. Ms. Fu responded that AI has significant potential to enhance the operation of complex systems like the electric grid by analyzing massive amounts of trustworthy data. She explained that DOE’s initiatives focus on using scientific data to train AI models, ensuring reliability in critical applications like grid management.Sen. King raised concerns about the U.S. losing talent due to a dysfunctional immigration system, noting that many foreign students now leave after completing their education. Dr. Kaushik confirmed that restrictive immigration policies contribute significantly to the loss of top talent, particularly from partner and allied nations. He cited Chinese policies that aim to attract high-end talent and emphasized the importance of balancing security with openness to retain international experts.Sen. King suggested offering green cards to graduates, assuming sufficient security measures are in place, to retain talented individuals who complete their education in the U.S. Dr. Kaushik agreed, highlighting the need for targeted security provisions rather than broad restrictions, which could allow the U.S. to attract and retain the global talent necessary for maintaining its technological edge.
Sen. Murkowski inquired about how AI and advanced technologies could be used in other areas, such as resource exploration, disaster prediction, and climate resilience. She referenced the recent deadly landslide in her hometown of Ketchikan and discussed the need for better monitoring and communication of environmental data. Ms. Fu highlighted several areas where AI could be impactful, including climate modeling, energy resilience for remote communities, and infrastructure monitoring. She emphasized the importance of making data AI-ready, as DOE is doing with permitting and NEPA data, to facilitate broader applications and improve decision-making.Sen. Murkowski mentioned challenges related to accessing public data from agencies like BLM and the need for better coordination among stakeholders. Ms. Fu responded that DOE is working to make vast amounts of data from climate, energy, and infrastructure sources accessible and usable for AI applications. She noted that integrating and organizing data across different fields is critical for addressing complex scientific problems and applying solutions to real-world situations.Sen. Murkowski raised concerns about aging military assets in Alaska facing coastal erosion and the need for long-term planning. Dr. Kaushik noted that federal agencies, such as NOAA and the National Labs, collect vast amounts of data that are underutilized. He suggested that federal initiatives, such as creating digital twins and automating scientific discovery through cloud labs, could enhance predictive modeling and resource exploration, which are typically not incentivized for private sector investment.Dr. Gleason emphasized the complexity of preparing data for AI applications, noting that raw data must be organized, labeled, and structured correctly to be valuable. He described how Oak Ridge National Lab is using AI to stitch together satellite images to map and analyze global infrastructure, predicting building materials and assessing disaster impacts. This approach allows for more effective disaster response by providing before-and-after comparisons of affected areas, helping direct emergency resources quickly and efficiently.
Sen. Hoeven inquired about the most relevant metrics used to compare AI capabilities between countries, such as the U.S. and China, and why these metrics matter. He also asked about managing AI security and the challenges associated with retaining talent trained in AI, particularly when other countries might attempt to recruit them. Ms. Fu responded that the Department of Energy is exploring different ways to measure AI capabilities beyond traditional metrics like computing power or financial investment. She noted that understanding the usefulness and effectiveness of AI models is crucial, and DOE is actively engaged in red-teaming efforts to assess AI models, including those with potential national security implications.Sen. Hoeven raised concerns about the security of U.S. AI research and the risk of losing talent to other countries. Ms. Fu emphasized that DOE is vigilant about security threats and maintains a managed research environment with strict access controls. She acknowledged that while the U.S. cannot control where individuals go after working in national labs, the unique resources and mission-driven work at DOE attract talent. However, she warned that this leadership is not guaranteed, as other companies and countries offer competitive salaries and opportunities.Sen. Hoeven asked if the labs are seeing people leave for other countries. Ms. Fu confirmed that while some talent leaves for industry, which is preferable if it remains within the U.S., retaining skilled workers is a constant challenge. Dr. Gleason echoed these concerns, highlighting the difficulty of recruiting domestic talent in STEM fields and the need to engage students early, even at the middle school level, to inspire future scientists and engineers. He stressed that many who stay at the labs are motivated by the mission rather than financial gain.Dr. Kaushik noted that no single metric provides a complete picture of AI leadership, as each metric serves different aspects of the field, such as research output, commercialization, or specific technological advancements. He pointed out that while the U.S. remains a leader in frontier AI models, China is rapidly advancing, as seen with companies like DeepSeek and Huawei developing competitive AI technologies. Dr. Kaushik emphasized that an aggregate of various metrics is necessary to fully understand global AI standings and the challenge for U.S. labs is to maintain top-tier talent to continue driving innovation.
Sen. Daines highlighted the success of the 2018 National Quantum Initiative Act, which spurred significant growth in quantum research and economic development in Montana. He noted that Montana now hosts over 50 companies and 1,200 employees dedicated to quantum and photonics, and is home to the Monarch Quantum Foundry at Montana State University and the Headwater Technology Hub, focusing on smart photonics. Sen. Daines emphasized the importance of reauthorizing the DOE Quantum Initiative to maintain U.S. leadership in quantum research and thanked Chairman Manchin for his support of the bipartisan DOE Quantum Leadership Act.
Chairman Manchin addressed the importance of avoiding duplication between the DOE and NSF, particularly in the implementation of the CHIPS and Science Act and other initiatives related to AI and emerging technologies. He emphasized that the goal is to ensure a balanced approach where each agency supports the other’s strengths without redundancy. The Chairman underscored the bipartisan consensus on the need to invest in emerging technologies while safeguarding research from foreign espionage, highlighting ongoing efforts to protect critical research within a democratic framework.Chairman Manchin concluded by stressing the need to leverage DOE’s assets effectively to avoid reinventing the wheel and duplicating efforts between agencies. He noted the importance of a “scalpel approach” to protect U.S. research from foreign espionage while allowing the best scientific minds to contribute to national advancements. The Chairman affirmed that the committee would continue to work towards a balanced and secure approach to supporting innovation in quantum, AI, and other critical technologies.

SPECIAL TOPICS

AI and Quantum Computing Leadership:

Members repeatedly focused on maintaining and advancing U.S. leadership in AI and quantum computing due to their critical roles in national security, economic prosperity, and technological innovation. Several speakers, including Sen. Daines and Ms. Fu, emphasized the need for continued investment, development, and collaboration between public and private sectors to stay ahead of global competitors, particularly China. AI and quantum computing are viewed as transformative technologies that could redefine the future of defense, cybersecurity, and economic power. Keeping the U.S. at the forefront of these technologies is essential for maintaining a strategic advantage.

Security of AI Research and Workforce Retention:

A major focus of the hearing was around concerns about safeguarding AI research from foreign espionage and the challenges of retaining talent within the U.S. The discussion highlighted the risk of trained professionals leaving for other countries, particularly adversaries like China, which could lead to critical knowledge and skills being exploited against U.S. interests. Protecting intellectual property and retaining top talent are crucial for maintaining national security and ensuring that technological innovations are not used against the U.S.

Energy Demands and Grid Management:

The growing energy demands of AI and data centers, especially in states like Nevada, were a significant concern. Discussions, led by Sen. Cortez Masto and others, focused on how to manage these demands, optimize the grid, and ensure that energy supply keeps pace with technological growth. Reliable energy supply is foundational for the operation of advanced computing technologies. Addressing grid management and ensuring energy efficiency are critical for supporting the expansion of AI and other high-demand technologies.

ADD TO THE NIMITZ NETWORK

Know someone else who would enjoy our updates? Feel free to forward them this email and have them subscribe here.

Blog