Advanced Micro Devices (AMD) has announced a significant technological breakthrough in partnership with OpenAI. The two companies will collaborate on a multi-year, multi-generation agreement to deploy 6 gigawatts of AMD GPUs, specifically the Instinct™ MI450 Series.
Details of the Agreement
The initial phase of this agreement will see the deployment of 1 gigawatt of the MI450 GPUs beginning in the second half of 2026. OpenAI aims to leverage this deployment for its next-generation AI infrastructure.
Strategic Partnership Benefits
Both AMD and OpenAI stand to gain from this collaboration, as it amplifies their respective strengths in high-performance computing and artificial intelligence. AMD’s leadership in high-performance chips will support OpenAI’s ambition for expansive AI deployment.
- Initial Deployment: 1 gigawatt in 2H 2026.
- Total Agreement: Up to 6 gigawatts of AMD GPUs.
- Milestone Shares: AMD will issue OpenAI warrants for up to 160 million shares tied to performance milestones.
Statements from Leadership
Dr. Lisa Su, AMD’s CEO, expressed enthusiasm for the partnership, highlighting its potential to enable major AI advancements and contribute to the ecosystem. Sam Altman, OpenAI’s CEO, affirmed the partnership as a crucial step toward harnessing AI’s full potential.
Financial Implications
This partnership is projected to generate tens of billions of dollars in revenue for AMD while supporting OpenAI’s infrastructure development needs. The agreement also emphasizes strategic alignment that could enhance shareholder value for both entities.
Next Steps
The companies will continue collaborating on product roadmaps and technical optimizations. Their joint efforts aim to meet the growing demands in AI and high-performance computing.
AMD is set to hold a teleconference to discuss this announcement. There will be an update on its third-quarter financial results on November 4, 2025.
Overall, this collaboration marks a pivotal moment in the advancement of AI technology, setting the stage for future innovations.
