OpenAI has more big deals coming after Stargate, Nvidia and AMD: Sam Altman
OpenAI has announced a string of mega‑agreements — from the Stargate venture with Oracle and SoftBank to multi‑billion‑dollar chip deals with Nvidia and AMD — and executives have indicated that further partnerships are on the way to meet soaring compute demand.
OpenAI chief executive Sam Altman has signalled, even after announcing the Stargate infrastructure venture and landmark partnerships with Nvidia, Oracle and AMD, that further major agreements are in the pipeline.
Recent OpenAI deals
Stargate
OpenAI unveiled “Stargate”, a joint venture with Oracle and SoftBank to build multiple US AI data‑centre campuses, with an initial $100 billion commitment and plans that could reach $500 billion over four years. The first mega‑site in Abilene, Texas has been showcased publicly alongside the partners.
Nvidia
OpenAI announced a staged partnership in which Nvidia has committed up to $100 billion, tied to building at least 10 GW of next‑generation AI capacity for OpenAI’s models. The deal has been presented as infrastructure financing and technology enablement.
AMD
OpenAI agreed a multi‑year chip‑supply deal for AMD’s forthcoming Instinct MI450, targeting the deployment of compute equivalent to 6 GW from 2026. The agreement has included a warrant that could allow OpenAI to acquire up to 160 million AMD shares (about 10%), subject to milestones.
Oracle
Beyond the cloud extension announced in 2024, Oracle has been expanding its role as a build‑out and leasing partner under the Stargate banner, with additional multi‑GW capacity disclosed during 2025.
Unprecedented scale
Across a whirlwind period in late September 2025, OpenAI has outlined plans amounting to 17 GW of AI infrastructure roughly the output of 17 nuclear reactors in partnership primarily with Oracle, Nvidia and SoftBank.
Altman has argued that even this pace will “look slow” as usage of ChatGPT and agentic systems has accelerated.
OpenAI has said it needs vastly more compute to advance frontier models and agents, and has therefore diversified beyond Nvidia to AMD while deepening ties with cloud and build partners.
OpenAI has pursued interlocking agreements that have combined procurement, financing and strategic alignment across chips, cloud and construction.
The approach has offered multiple supply lanes for compute while spreading risk and has tied OpenAI’s success to that of key suppliers and investors.