Nvidia to Invest Up to $100B in OpenAI (nvidianews.nvidia.com)

NVIDIA and OpenAI signed a letter of intent for a landmark strategic partnership in which NVIDIA will enable OpenAI to build and deploy at least 10 gigawatts of AI data-center capacity using NVIDIA systems — reportedly representing millions of GPUs — and will invest up to $100 billion progressively as each gigawatt is brought online. The first gigawatt of NVIDIA hardware is targeted for the second half of 2026 on NVIDIA’s Vera Rubin platform. OpenAI will designate NVIDIA as a preferred compute and networking partner and the two companies will co‑optimize model/infrastructure and hardware/software roadmaps as they scale. Technically, this signals a major acceleration of the compute arms race: 10 GW of dedicated AI capacity implies massive power, cooling and data‑center builds and enables far larger model training runs and inference fleets, accelerating OpenAI’s push toward next‑generation models and its stated path “to deploying superintelligence.” Co‑optimization between model stacks and NVIDIA hardware could deliver step changes in throughput and efficiency, but it also concentrates enormous compute, commercial influence, and grid impact in a tight vendor relationship. The pact complements OpenAI’s broader infrastructure partners (Microsoft, Oracle, SoftBank, Stargate) and will have wide implications for competition, supply chains, energy planning and AI governance as final terms are worked out.