(Bloomberg) -- For more than a year, Courtney Gibbons was focused on exploring the mathematical foundations of artificial intelligence — the kind of arcane research that can sometimes be overlooked until it helps pave the way for the next ChatGPT.
But in February, Gibbons was one of 170 employees fired at the National Science Foundation, a federal agency that has long been a linchpin of domestic technology research and investment. Many of these were probationary staffers and part-time experts like Gibbons who had been handpicked over the past two years specifically for their expertise in AI. About a quarter worked inside groups central to deploying NSF funding for AI research, according to documents viewed by Bloomberg.The wave of layoffs pushed by the Trump administration, combined with looming budget cuts, is now threatening the ability of the NSF to sustain AI research at the scale necessary for the US to remain competitive, according to industry watchers and current and former employees at the organization. These moves risk severing the talent pipeline that feeds the industry’s most cutting-edge companies and ceding leadership in artificial intelligence to China at a time when President Donald Trump has made it a priority to bolster US AI supremacy, largely through a deregulatory agenda.“This directly contradicts other Trump administration priorities,” said Gregory Allen, director of the Wadhwani AI Center at the Center for Strategic and International Studies. “Almost every employee with an advanced degree at every American AI firm has been a part of NSF-funded research at some point in their career. Cutting those grants is robbing the future to pay the present.”
On Monday, the agency said it would reinstate 84 of the workers to comply with a court ruling that determined the mass firing of probationary employees was illegal. The experts are not among those being being brought back, the NSF said. Attorneys for federal workers also say the temporary order could still be struck down.
A spokesperson for the NSF previously told Bloomberg News that the agency is “working expeditiously to conduct a comprehensive review of our projects, programs and activities to be compliant with the existing executive orders.” A White House spokesperson declined to comment.
One of the affected departments inside NSF, called the Directorate for Technology, Innovation and Partnerships, had been created as part of the 2022 Chips and Science Act and was considered an important avenue to funnel grants focused on machine learning, robots and advanced manufacturing. The group had been grappling with cuts to its funding, which was slashed by 30% in 2024 to $617.9 million. Now, layoffs mean there are fewer people inside the NSF who can ensure those grants are actually awarded.The layoffs hit AI-focused groups that were operating at capacity to begin with, according to current and former staffers. As a result, many review panels have been postponed or cancelled, stalling funding for some AI projects. It’s also left researchers and institutions that were already approved uncertain about who will be shepherding their projects in the months ahead. And the specter of additional cuts now hangs over academics across the country.
The newly formed Department of Government Efficiency, conceived by Trump and billionaire Elon Musk to slash federal spending, has given agencies including the NSF until March 13 to propose additional reductions that could reduce its workforce by up to 50%. Meanwhile, some Republicans in Congress are setting their sights on even more drastic cutbacks. America’s innovation engineCreated in 1950, the NSF has long been a quiet powerhouse for American innovation. The technology behind Larry Page’s and Sergey Brin’s PageRank algorithm, which helped sort digital information and served as the basis for Google, emerged from a project funded by the NSF.Years before generative artificial intelligence went mainstream with the launch of ChatGPT, the NSF also helped lay the foundations for the underlying technology behind AI chatbots.“It’s the place that makes most of the hardcore AI research happen,” said Gibbons, the former NSF staffer and a professor at Hamilton College in New York. “When you think about who funded the work that led to transformers, when people thought neural networks were kind of crazy back in the day, that was coming out of this group.”
Much of the conversation around AI competitiveness is currently devoted to private US companies, such as OpenAI, Alphabet Inc.’s Google and Meta Platforms Inc., which are building cutting-edge artificial intelligence systems. But the programs that the NSF funds have long been the fertile ground from which America’s tech giants recruit, said Kumar Garg, president at Renaissance Philanthropy, a nonprofit advisory firm that connects wealthy individuals and foundations with science, technology and innovation projects. “They might write that seminal paper when they’re at Google, but NSF played an important part of their story,” he said.The NSF cuts, while severe, represent a continuation of a longer-term decline in US public research investment as a percentage of GDP. “Private R&D has picked up some of that slack but that’s not going to be enough,” said Garg, who served in the Obama administration’s office of science and technology. “China’s going in the other direction.”Days after the NSF cuts, Meta’s Chief AI Scientist Yann LeCun wrote in a post on LinkedIn that “the US seems set on destroying its public research funding system.” Most other tech leaders have remained silent.
An uncertain eraFederal scrutiny of the NSF only appears to be growing. A report led by Republican Senator Ted Cruz flagged $2 billion in NSF-funded projects for potential cuts. He zeroed in on funding descriptions that referenced diversity, equity and inclusion, even when the actual grants were earmarked for technical projects like computing infrastructure.
One such program was the $29 million DeltaAI project at the University of Illinois Urbana-Champaign which uses NSF funding to buy advanced computing chips from Nvidia Corp. to run AI workloads for AI research at institutions across the country. Similar chips-focused programs at San Diego State University and the University of Alaska Fairbanks were also flagged. Bill Gropp, director of the National Center for Supercomputing Applications at the University of Illinois, said there appeared to be confusion over what the program means by diversity. “When we’re talking about diversity of researchers, we’re talking about people from different fields, states and disciplines,” he said.So far, there’s no indication that the program’s budget is being cut, he said, but “it’s hard to know what happens next.”
One notable alumnus of the University of Illinois and a beneficiary of its computing resources is venture capitalist Marc Andreessen. In a January interview, he said his research there resulted in the Mosaic web browser that helped popularize the internet. Now, he's helping support DOGE effort, recruiting candidates and calling himself the group's “unpaid intern” in an interview last month with the Hoover Institution.With government support uncertain, some researchers and organizations are considering alternative funding sources, including through philanthropies. But there are limits. “Philanthropic funding of the sciences is often aimed at a few schools,” said Margaret Martonosi, a professor at Princeton University who previously served as head of the NSF’s directorate for Computer and Information Science and Engineering. “It doesn’t help an aspiring AI expert in an arbitrary part of our country get the opportunities they need.”
