AMD and OpenAI have forged a landmark 6-gigawatt agreement, aiming to power the next-generation AI infrastructure via successive generations of AMD Instinct GPUs. Initial deployment, using 1 gigawatt of AMD Instinct MI450 GPUs, is expected to commence in the latter half of 2026. This ambitious project positions AMD's high-performance computing systems alongside OpenAI’s innovative AI research, highlighting their vital role in pivotal AI advancements.
According to the agreement, OpenAI will bolster its strategic compute partnership with AMD, focusing on large-scale deployments, starting with the AMD Instinct MI450 series. This extends to future generations and rack-scale AI solutions, building upon the existing collaboration that began with the MI300X series and continued with the MI350X series. Together, AMD and OpenAI will create a beneficial ecosystem, ensuring extensive AI deployments on a massive scale.
A substantial part of this agreement includes AMD issuing OpenAI a warrant for up to 160 million shares of AMD common stock. This structure involves vesting upon achieving particular milestones, with the initial 1-gigawatt deployment prompting the first tranche. Securement of additional tranches hinges on purchasing scale-ups to 6 gigawatts, dependent on AMD's share price targets and OpenAI's technical and commercial achievements aligning with the broader deployment aims.
Dr. Lisa Su, Chair and CEO of AMD, highlighted the significance of this partnership, saying the collaboration will lead to the world’s most ambitious AI projects, benefitting the overall AI ecosystem. Similarly, OpenAI’s CEO, Sam Altman, emphasised the major strides in computing capacity necessary to reach AI's full potential, with OpenAI's President, Greg Brockman, pointing out the importance of collaboration across every technology stack layer. AMD's EVP, Jean Hu, stated that this partnership could generate substantial revenue for AMD while accelerating OpenAI's infrastructure.
The alliance signifies a strategic melding of AMD's and OpenAI’s resources and expertise to meet escalating global demands for AI infrastructure. The combination of innovative strategies and effective execution is anticipated to expedite the future of high-performance computing and AI solutions.
