Trump Administration Secures $100 Million AI Compute Credits as National Compute Targets 2 Gigawatts by 2030
Updated
Updated · POLITICO · Oct 7
Trump Administration Secures $100 Million AI Compute Credits as National Compute Targets 2 Gigawatts by 2030
1 articles · Updated · POLITICO · Oct 7
Summary
$100 million in compute credits will go to a Trump administration AI science initiative, giving researchers access to a new shared computing grid the White House declined to discuss.
The network is designed to pool capacity from labs and cloud providers using Nvidia and AMD chips and Google TPUs, letting users buy compute on demand, reserve larger blocks or connect their own systems while paying only for power used.
National Compute, the company behind the grid, says it has allocated 750 megawatts so far and aims to assemble 2 gigawatts of capacity by 2030, backed by chipmakers, infrastructure firms, universities and institutional investors.
A parallel "Grid Defense" program will let government agencies test AI models through an embedded agent called Marshall, with agencies setting confidential evaluation rules and receiving results before deciding on deployment.
The initiative fits the administration's light-touch AI strategy, which emphasizes widening compute access and accelerating U.S. model development in competition with China.