In a major move for the artificial intelligence industry, AMD has signed a multi-year chip supply deal with OpenAI, aimed at providing up to 6 gigawatts of compute capacity across multiple generations of its Instinct GPUs. The agreement is expected to generate tens of billions in revenue for the chipmaker while accelerating OpenAI’s AI infrastructure expansion.
A Massive AI Compute Deployment
Under the deal, AMD will supply 6 gigawatts of computing power to OpenAI enough to power approximately 4.5 million homes. The deployment will begin with AMD’s Instinct MI450 GPU, with the first gigawatt slated for the second half of 2026. The partnership will extend to subsequent generations of AMD’s GPU offerings, incorporating improvements in both hardware and software, many of which will be informed by OpenAI’s input.
The MI450 series is positioned to compete directly with Nvidia’s Rubin CPX GPUs, offering performance enhancements designed for large-scale AI workloads. AMD’s existing MI355X and MI300X GPUs, which are already used by OpenAI for some AI inference tasks, are renowned for their high memory capacity and bandwidth, making them suitable for large language model computations.
Equity and Financial Stakes
As part of the agreement, OpenAI has been granted the option to purchase up to 160 million shares of AMD stock, representing a 10% stake in the company. The vesting of this stake is tied to the incremental deployment of compute capacity, starting with the initial 1 gigawatt and scaling up as OpenAI acquires the full 6 gigawatts. Additional tranches are tied to AMD stock price milestones, with the final tranche vesting if the stock reaches $600 per share. Following the announcement, AMD shares jumped nearly 35%, opening at $222.24 after closing at $164.67 the previous Friday.
Strengthening AI Infrastructure
This strategic partnership comes as OpenAI races to secure chip supply for its expanding AI infrastructure. The company’s Stargate initiative, which includes collaborations with Oracle, SoftBank, Samsung Electronics, and SK Hynix will expand data center capacity with a planned 7 gigawatts of compute. These agreements collectively position OpenAI to scale AI compute infrastructure at an unprecedented level.
OpenAI CEO Sam Altman described the AMD partnership as “a major step in building the compute capacity to realize AI’s full potential,” highlighting the importance of diversified hardware partnerships in the AI race. AMD’s CEO Dr. Lisa Su echoed this sentiment, noting that the collaboration brings together the best of both companies to advance the entire AI ecosystem.
A Competitive AI Chip Landscape
The AMD deal follows a series of major partnerships that OpenAI has secured in recent weeks. Nvidia agreed to invest up to $100 billion in OpenAI while supplying at least 10 gigawatts of compute, and OpenAI signed a $10 billion agreement with Broadcom for custom AI chip development. Meanwhile, collaborations with Samsung and SK Hynix aim to provide the DRAM memory necessary for AI data centers in South Korea.
These strategic moves demonstrate OpenAI’s aggressive expansion strategy and its commitment to building robust, large-scale AI infrastructure, positioning the company as a central player in the rapidly evolving AI ecosystem.
