Trump’s Return to Office Sparks Focus on AI Infrastructure

Artificial intelligence has been such a prominent focus in the federal government over the past year that officials have been eyeing what that means for the incoming administration next month.

“As the administrations have handed the baton to each other, even though the political parties have changed, the overall emphasis on investing more in AI and seeing AI as a key emerging technology has only increased,” Kumar Garg, former senior advisor to the deputy director at the White House Office of Science and Technology Policy (OSTP), told GovCIO Media & Research.

What Will Trump Do For AI?

Donald Trump in his first term had named AI as a research and development priority in 2017. In 2019 he signed an executive order dubbed the American AI Initiative, a national strategy aimed at promoting U.S. dominance in AI research, development and deployment. This was followed by a December 2020 executive order promoting trustworthy AI.

President Joe Biden’s 2023 AI executive order focused on developing standards focused on AI safety, security and trust.

The 2024 Republican Party Platform criticized Biden’s executive order on AI for what it perceived as creating barriers to innovation. “Republicans support AI development rooted in free speech and human flourishing,” the document said.

Regardless of how the new administration decides to approach policymaking around AI, Garg said investments in the U.S. energy infrastructure and R&D will be critical to the nation’s competitive advantage.

As agencies have worked over the past year to complete some of the initiatives from the executive order, including naming chief AI officers, some officials point to the commonalities between the two administrations that could drive future AI efforts.

The bipartisan House Task Force on Artificial Intelligence outlined what a politically shared vision for AI looks like in its newest report released this month. The report outlined numerous key findings and recommendations about standards development while also emphasizing the value of the private sector.

“Despite the wide spectrum of political views of members on our task force, we created a report that reflects our shared vision for a future where we protect people and champion American innovation. We have made our best efforts based on the information we have, but with the rapid pace of change in both AI software and hardware, we are fully aware that we don’t know what we don’t know. This initial report is only the first step,” said the task force’s co-chair Rep. Ted Lieu about the report.

Building the Energy Infrastructure to Power AI

The growing demand in AI means electrical demand to power it is also increasing, presenting challenges for physical infrastructure availability, electrical grid reliability and affordable electricity. Department of Energy (DOE) officials said electricity consumption in the U.S. has grown at 0.5% per year in the last two decades, with recent estimates suggesting an annual growth of at least 0.9% until 2030 and an increase in the five-year cumulative growth forecast from 2.6% to 4.7% due to the surge of AI and data centers.

“A new, emergent issue is this intersection between AI and energy. Can we build out enough energy capacity to sustain [AI]? If the scaling laws hold, and the leading firms continue to build out large need for compute and training, what kind of data center capacity will be needed? And how much will that put pressure on the grid? And where can we build?” Garg said.

Garg noted the second Trump administration would face two key challenges in building out energy capacity to support AI: permitting and transmission.

“For example, we’ve seen in states like Texas and others that have [built] regulatory policies … and outstrip in clean energy deployment,” Garg said. “Then comes things like transmission, which is if you’re building in one place and you want to move to another … that might create a little bit of a race between the states to say, ‘If we don’t build a relevant energy capacity, folks are not going to build their data centers in our state.’”

Innovations in the energy sector are essential to the future of AI. Analysts from the Special Competitive Studies Project (SCSP), a think tank based in Arlington, Virginia, noted the role of the energy sector for U.S. dominance in AI.

“China’s aggressive pursuit of renewable energy technologies and dominance in critical mineral supply chains presents a direct challenge to U.S. leadership in the energy sector,” said its recent memo.

It outlined five key obstacles: insufficient supply of existing energy sources to power AI, regulatory hurdles that create barriers to innovation, physical infrastructure limitations of the U.S. electrical grid, market fragmentation that limits collaboration and global competition from foreign adversaries.

Some of its recommendations include streamlining regulatory processes, securing the grid and fostering partnerships.

The Biden administration during a September roundtable launched several new actions targeted at the energy infrastructure, including a new task force on AI data center infrastructure to coordinate policy across government, scaling up technical assistance for data center permitting and creating an AI data center engagement team within DOE.

The Congressional task force’s AI report recommended federal leaders support and increase investments in scientific research that enable energy infrastructure, strengthen efforts to track and project AI data center power use, and ensure that AI and the energy grid are a part of broader discussions about grid modernization and security.

Another matter in the increased needs from the power grid includes ensuring it’s cybersecure.

“The DOE, the Department of Homeland Security (DHS), the North American Electric Reliability Corporation (NERC), FERC, and the Electricity Information Sharing and Analysis Center (E-ISAC), should implement comprehensive and advanced cybersecurity measures, such as real-time threat monitoring, to protect the entire energy system — from generation to distribution — from cyberattacks,” said SCSP’s report.

Last year, DOE and the Cybersecurity and Infrastructure Security Agency (CISA) released new measures to ensure security and resilience of the power grid in response to various incidents throughout 2022 that threatened electrical substations throughout the U.S.

“While the work has always been ongoing, it was critical to bring it back to the attention of everyone and remind them of those best practices,” DOE Preparedness, Policy and Risk Analysis Deputy Director Mara Winn told GovCIO Media & Research in a 2023 interview. “We wanted to make sure there were resources to guide the awareness of the threat environment, what implementation of protective physical security measures were possible and have that layered security strategy to ultimately reduce or minimize the impact of an attack.”

Investing in AI R&D

Public sector leaders agree that research and development is essential for the U.S. to stay competitive in AI.

“Maintaining strategic leadership on AI has been pretty consistent. Then the question becomes what are the big parts of that? Part of that is R&D. So if there are cuts to the U.S. investments in R&D, I think that would be quite negative,” said Garg. “Some of these other blockers that are popping up, like energy — [the Trump administration] has already said that they’re going to make sure that there are strong federal partnerships with the emerging industry.”

Under Biden’s AI executive order, federal agencies are required to inventory AI use cases and share them across government and with the public. Earlier this month, agencies published an updated inventory that listed over 1,700 use cases.

That National Science Foundation (NSF) has been at the forefront of AI R&D with 16 published AI use cases in the inventory, including AI voiceover, FOIA processing and its “Ask NSF Initial Pilot.”

As part of Biden’s executive order, NSF launched the National AI Research Resource (NAIRR) pilot program in partnership with 10 other federal agencies and 25 non-governmental partners to make additional resources available to support the nation’s research and education community. The agency intends to institute the full program in 2026.

Dorothy Aronson, chief AI officer at NSF, told GovCIO Media & Research the pilot activities are laying the groundwork for eventual full-scale AI research infrastructure. So far more than 200 resource awards have been made to support AI innovation and cybersecurity in agriculture, health care and wildfire detection.

“The AI breakthroughs we are seeing are the result of many years of investment in AI research and education, in which NSF played a critical role,” Aronson told GovCIO Media & Research. “Development and education are just as important, as the benefits of AI can only be shared broadly if many people are informed and aware, and the right methods are developed in ways that solve important problems.”

Aronson said listening to the researchers on the frontline is the most important method NSF uses to stay ahead in research and development. As the U.S. tries to maintain its edge with AI, trends and feedback from the research community will influence the development of new programs and funding opportunities.

“Competing with China will likely remain front and center. I think that there’ll be much more of a focus in [Trump’s] administration around research and development funding through agencies like DARPA [and] NSF,” Bill Wright, who had served as senior operations officer at the Office of the Director of National Intelligence and now leads government affairs at Elastic, told GovCIO Media & Research in an interview.

What’s Next for AI Policy?

Amid Trump’s intentions to counter Biden’s approach to AI, he recently tapped former chief operating officer at PayPal David Sacks as his White House AI and crypto czar.

Garg’s perspective is that leaders should develop a flexible approach and focus on improving forecasting to drive data-driven decision-making across government, building institutions to manage AI, and recruiting and retaining a top AI talent pipeline.

“The best way to think about the regulatory framework is not [that] you’re trying to manage for today. What you’re trying to do is say, ‘Okay, do we have the talent in place? Do we have the cooperative models between industry and government in place and others so that, as the models keep getting more powerful, you’re able to figure it out?’” said Garg. “It’s very hard to create these bright-line rules because very often with a rapidly evolving technology, the rules end up being brittle. So you need to build more of a flexible approach.”