OpenAI Strikes Major Chip Deal with AMD, Securing Future AI Compute
Key Points
- OpenAI will purchase six gigawatts of AMD Instinct MI450 GPUs over several years.
- The deal includes an option for OpenAI to acquire roughly a ten‑percent stake in AMD.
- First gigawatt of AMD GPUs is slated for deployment in the second half of 2026.
- OpenAI will continue to expand its use of Nvidia GPUs alongside AMD hardware.
- AMD calls the partnership a breakthrough and expects it to attract more tier‑one customers.
- The agreement aligns with a U.S. initiative to invest billions in AI data‑center infrastructure.
- Industry analysts cite scaling laws as a key driver of massive AI compute demand.
OpenAI announced a multi‑year agreement to purchase a massive volume of AMD chips, including the right to acquire roughly a 10 percent stake in the chipmaker. The deal commits OpenAI to buying six gigawatts of AMD Instinct MI450 GPUs over the coming years, with the first gigawatt slated for deployment in the second half of 2026. In addition to expanding its AMD partnership, OpenAI will continue to invest in Nvidia hardware. The agreement is positioned as a key component of a broader U.S. push to build AI data centers and reflects the growing demand for compute power across the generative‑AI ecosystem.
Deal Overview
OpenAI disclosed a landmark agreement to acquire a substantial quantity of AMD chips for its data‑center operations. The company will purchase six gigawatts of AMD Instinct MI450 GPUs over the next several years, with the first gigawatt of hardware expected to be installed in the second half of 2026. The contract also grants OpenAI the option to buy up to 160 million shares of AMD common stock, which would represent roughly a ten‑percent ownership stake. Alongside the AMD commitment, OpenAI confirmed it will continue to expand its use of Nvidia GPUs, underscoring a diversified hardware strategy.
Strategic Implications
AMD hailed the partnership as a breakthrough achievement, describing OpenAI as a strategic supplier that could attract additional tier‑one customers. The deal builds on a pre‑existing relationship; at AMD’s Advancing AI event, OpenAI CEO Sam Altman joined AMD CEO Lisa Su on stage, noting that AMD had been designing the MI400 series with feedback from marquee customers like OpenAI. Su referred to Altman as a “great friend” and an “icon in AI.” Altman emphasized the need for “tons of compute, tons of memory, and tons of CPUs” in addition to Nvidia GPUs, highlighting the breadth of hardware required for modern generative‑AI models.
Industry Context
The agreement arrives amid a broader U.S. initiative to invest heavily in AI data‑center infrastructure. The president announced a collaborative effort involving OpenAI, SoftBank, and Oracle to fund large‑scale AI data‑center projects on American soil. Analysts note that the data‑center “gold rush” is driven by the belief that AI model performance scales with increased compute and data. Industry forecasts project AI infrastructure spending at around $400 billion, while consumer‑facing AI services generate roughly $12 billion in revenue. The OpenAI‑AMD deal is therefore positioned as a critical component of the nation’s push to secure the compute capacity needed for the next generation of AI applications.