Overview
- The Office of Science and Technology Policy issued guidance this week proposing to move federal R&D support away from institution‑mediated grants and toward individual scientists and AI‑centered projects, a change that would affect about $200 billion in annual research allocations.
- The White House announced a $5 billion AI science initiative that will pool Department of Energy supercomputers and cross‑agency datasets to speed drug discovery and study chronic disease, with Microsoft pledging compute credits to the program.
- OpenAI disclosed plans for a major self‑led data center in Effingham County, Georgia with roughly $20 billion in upfront commitments and said it has substantially raised its pre‑2030 compute spending forecast, signaling far larger long‑term capital needs for training models.
- Industry deals are locking in chips and capacity: AMD said it will make up to a $5 billion strategic investment in Anthropic while Anthropic agreed to procure up to 2GW of AMD’s next‑generation AI chips, and Alphabet and Tesla have raised capex guidance to expand their own AI infrastructure.
- The combined public and private push tightens a global race for compute, data and chips that raises legal and governance questions, pressures university funding models, and creates local impacts from energy and water use to job and community benefit negotiations.