OpenAI and AMD seal multi-year partnership to power 6 GW of AI compute

AMD and OpenAI revealed a strategic partnership to co-develop and deploy AI computing capacity at an unprecedented scale. The 6 GW agreement, spanning multiple generations of AMD Instinct GPUs, positions both companies as central players in the next wave of AI acceleration.

The first deployment, 1 GW of AMD Instinct MI450 GPUs, will begin in the second half of 2026, expanding through subsequent generations. AMD’s history in high-performance computing and OpenAI’s breakthroughs in generative AI make this a powerful alliance at a crucial stage in AI’s evolution.

Aligning strategies and incentives

Under the terms of the partnership, AMD becomes a core compute partner for OpenAI, integrating rack-scale AI systems optimised for OpenAI’s growing model workloads. To cement long-term alignment, AMD has issued OpenAI a warrant for up to 160 million AMD shares, tied to deployment milestones and AMD share price targets.

The first tranche of shares vests upon the 1 GW rollout, while further tranches unlock as deployments scale toward the 6 GW target. Vesting is also linked to OpenAI achieving technical and commercial objectives necessary for large-scale AMD integration.

Leadership vision

Dr Lisa Su, AMD’s Chair and CEO, said the partnership will “enable the world’s most ambitious AI buildout,” emphasising AMD’s expanding role in AI infrastructure.

Sam Altman, OpenAI’s co-founder and CEO, echoed the sentiment: “AMD’s leadership in high-performance chips will enable us to accelerate progress and bring the benefits of advanced AI to everyone faster.”

A long-term roadmap for AI growth

This partnership builds upon a history of collaboration between AMD and OpenAI, which began with the MI300X GPUs and evolved through the MI350X series. By jointly optimising hardware and software roadmaps, the companies aim to deliver more efficient, scalable compute for the next generation of AI models.

AMD and OpenAI’s multi-year deal highlights the intensifying race to secure GPU capacity for large language models and generative AI platforms. With 6 GW of GPU power planned, the scale rivals national data centre infrastructure projects, signalling how central compute availability has become to AI’s future.

Industry implications

For AMD, this partnership marks another milestone in its competition with Nvidia, the long-dominant supplier in AI accelerators. For OpenAI, it diversifies supply sources and helps future-proof its capacity needs as demand for ChatGPT and other AI services continues to grow.

Together, AMD and OpenAI are betting big on the future of AI compute, laying the groundwork for what could become one of the most significant technology collaborations of the decade.