National Compute announced Thursday that it’s donating $100 million in computing credits to the Trump administration’s Genesis mission, a government-wide initiative that uses AI to accelerate scientific discovery. The commitment signals confidence in the administration’s market-driven approach to AI innovation and removes a significant barrier for researchers who struggle to access expensive computing infrastructure.
White House Office of Science and Technology Policy Director Michael Kratsios is set to receive the announcement at a Thursday event celebrating what administration officials call a “golden age of science.” The credits will give Genesis Mission researchers access to advanced computing resources. Smaller labs and companies can struggle to secure capacity for frontier research, while major providers often prioritize long-term contracts with the largest AI labs.
A Market Solution to Computing Scarcity
National Compute, a newly formed company that’s pooled computing power from major technology providers, is building what it calls the National Compute Grid. The network will aggregate computing capacity from multiple sources and make it available to researchers, companies, and government agencies—a distributed marketplace instead of government-directed allocation.
The grid draws computing capacity from labs and cloud providers deploying Nvidia and AMD processors, as well as Google’s tensor processing units. Users can purchase computing as needed, reserve capacity for larger projects, or connect their own systems. They pay only for the power they consume. This usage-based model contrasts sharply with centralized government resource allocation and allows efficient pricing of scarce resources.
National Compute has backing from major chipmakers, cloud and infrastructure providers including Vultr and Crusoe, and universities. Venture capitalists and institutional investors finance the company, including public investors such as universities and pension funds. It aims to pool 2 gigawatts of computing capacity by 2030, with 750 megawatts allocated to date.
“We think there should be thousands more Anthropics and OpenAI’s,” said National Compute Cofounder Anjney Midha. “But to unleash that kind of innovation, which is extraordinarily profitable for the public, we need to reduce the barrier to entry, increase compute access, and allow the researchers and scientists to do what they do best.”
Voluntary Testing, Not Mandates
National Compute is also launching a “Grid Defense” program that lets government agencies test AI models through an embedded agent called Marshall before deciding whether models are safe enough to deploy. Agencies set their own testing procedures instead of following uniform federal requirements. The government can see the results, while labs can decide voluntarily whether to release a model.
The program entered invite-only early access in June with some government agencies. Starting this week, National Compute is opening access to public-sector users with .gov or .mil email addresses, as well as students and faculty with .edu addresses. Each public-sector user gets access to $100 in National Compute credits, in addition to the $100 million donation for the Genesis mission.
Midha described the approach as “continuous frontier alignment monitoring at scale” that would “enable your public stakeholders to monitor you as they need.” The program depends on buy-in from AI labs and the federal government, not regulatory coercion. National Compute created it partly in response to high-profile incidents involving AI agents, including attacks on companies that AI labs reported over the summer.
The Trump administration’s light-touch regulatory approach prioritizes winning the race with China to build the most advanced AI models. Instead of heavy-handed restrictions, the strategy emphasizes removing barriers to private innovation while using transparent testing mechanisms that respect voluntary industry participation.
This public-private partnership model shows how markets can solve infrastructure problems without government ownership or operation of the underlying systems. Private companies competing to provide computing services create efficiency incentives that centralized government provision typically lacks.
Why This Matters:
The $100 million commitment reveals how private capital and market mechanisms are stepping in to address genuine bottlenecks in AI research infrastructure. Instead of government building and operating computing systems, industry is pooling resources voluntarily because it’s profitable and aligns with its interests. The Grid Defense model also shows how voluntary, agency-directed testing can replace mandated federal compliance frameworks while maintaining accountability. This approach respects both innovation and safety without bureaucratic overhead. The stakes are genuine: the administration explicitly frames this as competition with China, making efficient resource allocation a matter of national interest. How government leverages this private-sector contribution, and whether the voluntary model proves sufficient for safety concerns, will determine whether market-driven AI governance can work at scale.