AMD to supply OpenAI with 6GW-worth of GPUs in multi-year deal — 10% stake option deepens the non-Nvidia silicon axis OpenAI is building alongside Broadcom Jalapeño custom chip
AMD will supply OpenAI with hundreds of thousands of GPUs in a multi-year deal — total power consumption 6GW, deployment starting 2026. OpenAI receives the option to take a 10% stake in AMD. Combined with the Broadcom Jalapeño custom chip and AMD GPU supply, OpenAI is structurally building a non-Nvidia silicon axis at substantial scale.
The substantive piece is the OpenAI non-Nvidia silicon strategy assembling at scale. Jalapeño announcement yesterday addressed the custom-silicon dimension. The AMD 6GW multi-year deal adds substantial GPU supply on the AMD silicon platform. The 10% stake option creates capital-architecture alignment alongside the supply commitment. The combined OpenAI compute strategy now substantially diversifies beyond Nvidia dependency.
The competitive read for Nvidia's structural compute-vendor position is that OpenAI explicitly moves to multi-vendor silicon supply. Through 2025 OpenAI was substantively Nvidia-dependent for frontier training and inference. The H2 2026 to 2030 OpenAI compute supply will operate on Nvidia + AMD + Broadcom custom + hyperscaler infrastructure (Stargate, Azure). Four-vendor compute supply substantially reduces Nvidia's structural pricing power over OpenAI procurement.
DCD — AMD to supply OpenAI with 6GW-worth of GPUs, 1GW deployment from 2026 → · GPU Insights — NVIDIA AMD AI chips in 2026: Blackwell, MI400 & Gaudi →