AMD Partners with OpenAI for AI Chip Supply
On Monday, AMD announced it has struck a deal to supply OpenAI with a significant number of AI chips, which also includes a 10% ownership stake in the company.
The news led to a 27% surge in AMD’s stock during the morning, boosting its market value by nearly $80 billion as investors recognized AMD as one of the primary computing partners for OpenAI.
This agreement outlines a substantial 6 gigawatts of power, sufficient to support around 5 million households in the U.S. during the latter half of 2026 and for several additional years.
AMD anticipates this partnership will yield hundreds of billions in revenue and accelerate the development of OpenAI’s AI infrastructure. OpenAI’s CEO, Sam Altman, commented that the collaboration is a crucial advancement in creating the computational capabilities needed to unlock the full potential of AI.
He emphasized that AMD’s expertise in high-performance chips will speed up the progress and make advanced AI benefits more accessible to everyone.
OpenAI has plans to establish a 1 gigawatt facility next year, utilizing AMD’s upcoming Instinct MI450 chips. As part of their agreement, AMD has given OpenAI a warrant for up to 160 million shares of common stock, contingent upon reaching specific milestones like the initial development of the 1 gigawatt facility.
If OpenAI fully exercises this option, it could end up with approximately a 10% stake in AMD.
AMD’s CEO, Lisa Su, expressed enthusiasm about teaming up with OpenAI to facilitate large-scale AI computing. She noted that the collaboration will leverage the strengths of both companies, fostering significant advancements in the AI landscape.
This contract stands out as one of the largest GPU deployment agreements, surfacing just two weeks after OpenAI disclosed a $100 billion chip deal with competitor Nvidia. The AMD partnership is expected to decrease OpenAI’s reliance on any single vendor, potentially disrupting Nvidia’s dominance in the sector; Nvidia shares fell by 1.1% on Monday following the announcement.
Additionally, OpenAI is in discussions with Broadcom to develop custom chips for the next generation of GPUs, reflecting broader trends toward a more intricate supply chain in AI.
In this competitive landscape, Nvidia provides the funding for chip purchases while Oracle assists in constructing large data centers to meet necessary power demands. AMD and Broadcom supply the GPUs, while OpenAI creates the market demand.
This scenario mirrors the Stargate Project, a $500 billion initiative announced by President Trump in January, aimed at rapidly establishing AI infrastructure in the U.S. involving OpenAI, Oracle, and SoftBank.
OpenAI’s initial data center under this project is already operational in Abilene, Texas, where it utilizes Nvidia chips, with construction progressing to expand its capabilities.
Future data centers are in the pipeline for New Mexico, Ohio, and the Midwest, likely depending on a mix of suppliers, including AMD.
Last month, OpenAI revealed it had reached an agreement to transition into a for-profit organization, despite facing legal challenges from Elon Musk, a rival to Altman.
