Public-private partnerships need more ‘efficiency,’ Energy official says

Leaders within the Department of Energy said on Monday they are poised to leverage the agency’s high performance computing resources to help advance U.S. leadership in artificial intelligence innovation, but they also moved to underscore the need for better processes to support faster industry partnerships.

Speaking at the AI + Expo in Washington, D.C., Harriet Kung — the acting director at the Department of Energy’s Office of Science — referenced the successful history the agency’s national lab apparatus has had with engaging industry partners to foster major scientific and technological innovation. 

Collaborations like these will be critical to leverage the best of Energy’s capabilities, but need to occur at an expedited rate, she noted. 

“We need new public-private partnerships,” she said. “We need speed, we need efficiency, we need mutual benefits. Traditional negotiations that we have had with private companies [have] been too slow, in my view, and that is something that we are eager to work with all industry partners to try to get to a better way to get the partnership going much faster.”

The labs have the ability to provide private companies, such as OpenAI or Anthropic, with well-curated data to help them effectively train their models, while simultaneously using this foundation to build a more science-focused AI tool. 

“We as an agency would have to really increase our funding and AI research, because we have to figure out how to construct and how to tune more of the science-based models,” she said. 

Kung added that hastening the pipeline to Energy’s national labs and private sector partnerships will accelerate breakthroughs, particularly by giving industry access to Energy’s high performance computing resources. 

“If we look at AI, it requires energy, it requires compute, and [at] DOE, we're well known for our leadership in computing and also in developing novel imaging technologies,” Kung said. “We think that public-private partnership absolutely has to be the answer to winning this race.”

Noting the U.S. national lab apparatus is outfitted with advanced computer systems geared towards supporting AI and machine learning systems’ hunger for large volumes of data, Kung said that this infrastructure and its expert workforce should be harnessed to further drive AI research and innovation. 

“These people are making sure that we can capitalize on the exoscale computing system that we have in labs, and making sure that we have the software and algorithm to make these capabilities available to the broad community. I think that is a very, very important asset for the whole community,” Kung said. 

She further noted that Energy has a dual role in supporting U.S ambition to become a global leader in AI technology: driving innovation in AI systems and leveraging AI applications and solutions across research domains. The innovation component will require the capitalization and integration of Energy’s existing expertise in high performance computing. 

“On one hand, we're developing AI technologies, and on the other hand, we're using AI for advanced science in our mission areas,” she said. “I think very few federal agencies really would have the opportunity, the capacities, the expertise to really advance AI from AI development to AI applications.”

Leadership based within the national labs echoed Kung’s sentiments, mainly noting that extensive public-private partnerships can help federal researchers benefit from larger investments.

“The Anthropics, the Googles, the OpenAIs, they are making a massive investment in a core capability that becomes a platform on which we can build our more bespoke capabilities through suitably structured partnerships,” Thomas Mason, the director of Los Alamos National Lab, said during a separate panel on Monday. He clarified that this doesn’t mean there shouldn’t be any public funding in science and technology research, particularly for research ambitions that don’t necessarily have an immediate commercial impact.

“It's a question of structuring the right sort of arrangements where we get the benefits of their expertise and investments,” Mason said. “I think we bring something to the table in terms of the expertise we have, the data that we have and the competencies that we have.”

Savannah River National Lab Director Johney Green raised the point that academia is another key partner to help further federal science and technology research priorities. 

“The labs have several close partnerships with various universities and academic partners, and they're bringing in fresh ideas … and approaches, and that interchange of ideas helps keep us at the cutting edge,” he said.