AMD has secured a landmark deal with OpenAI that positions it as a major player in the AI infrastructure race. The partnership includes a commitment to deliver up to 6 gigawatts of AMD Instinct GPUs, starting with 1 gigawatt in late 2026, and grants OpenAI a 10% equity stake in AMD through a special share agreement.
Key Highlights
- Compute Deployment: OpenAI will use AMD’s upcoming Instinct MI450 and future MI500 GPUs to power its next-gen AI models and data centers.
- Equity Agreement: OpenAI receives 160 million AMD shares at $0.01 each, representing up to 10% ownership, contingent on deployment milestones.
- Diversification Strategy: This move reduces OpenAI’s reliance on Nvidia’s H100 and B100 GPUs and complements its partnership with Broadcom for custom AI chips.
Strategic Implications
- AMD’s ROCm stack becomes a core part of OpenAI’s software infrastructure, alongside PyTorch and Triton.
- Nvidia’s dominance in AI compute faces new pressure, especially as AMD scales its datacenter footprint.
- Broadcom, AMD, and Nvidia now form a three-way supply chain for OpenAI’s massive AI workloads.
Industry Reaction
- Analysts view this as AMD’s most aggressive move into the AI datacenter space.
- The deal was quietly revealed during AMD’s July 2025 launch event, with Lisa Su and Sam Altman appearing together on stage.
- OpenAI’s investment signals long-term confidence in AMD’s hardware roadmap and software stack.
AMD’s partnership with OpenAI marks a pivotal moment in the AI hardware race. With up to 6GW of Instinct GPUs committed and a 10% equity stake granted, AMD is no longer just a competitor—it’s a cornerstone of OpenAI’s future infrastructure. This deal signals a major shift in power dynamics, challenging Nvidia’s dominance and accelerating AMD’s rise in the AI datacenter space.