OpenAI has signed a partnership with semiconductor manufacturer AMD to supply high-performance chips for its AI infrastructure, with OpenAI also receiving the option to purchase up to 10% of AMD’s shares. The deal positions AMD as a key alternative to Nvidia’s dominance in AI computing while highlighting the massive infrastructure demands needed to support advanced AI development.
Key details: The agreement centers on AMD’s next-generation Instinct MI450 graphics chips, expected to debut in 2026.
• OpenAI will purchase chips capable of delivering 6 gigawatts of computing power for its “next generation” AI infrastructure.
• The first 1-gigawatt batch is scheduled for deployment in the second half of 2026.
• AMD issued OpenAI a warrant to buy up to 160 million shares of common stock, representing approximately 10% of the company based on AMD’s 1.6 billion outstanding shares.
Market impact: AMD’s stock surged 25% in pre-market trading following the announcement, while Nvidia shares declined slightly.
• The deal provides a significant boost to Santa Clara, California-based AMD, which has lagged behind rival Nvidia in the AI chip market.
• Barclays analysts noted the partnership demonstrates the “desperate” need for more computing power across the AI ecosystem rather than direct competition with Nvidia.
Strategic context: The partnership reflects OpenAI’s efforts to diversify its chip supply chain beyond Nvidia’s market-leading position.
• Last month, OpenAI announced a separate $100 billion partnership with Nvidia that will add at least 10 gigawatts of data center computing power.
• OpenAI has already installed hundreds of Nvidia’s GB200 computing racks at its flagship Stargate data center campus under construction in Abilene, Texas.
What they’re saying: “This partnership is a major step in building the compute capacity needed to realize AI’s full potential,” OpenAI CEO Sam Altman said in a news release.
• “AMD’s leadership in high-performance chips will enable us to accelerate progress and bring the benefits of advanced AI to everyone faster.”
The big picture: The warrant’s vesting depends on two milestones tied to computing power deployment and unspecified share-price targets, creating a performance-based investment structure that aligns both companies’ interests in scaling AI infrastructure rapidly.