OpenAI to Deploy 6 Gigawatts of AMD GPUs, Anchored on Instinct MI450 Series from 2026
In a move that cements its position at the forefront of AI infrastructure innovation, AMD has announced a definitive strategic partnership with OpenAI to power the next generation of global AI systems. Under the agreement, OpenAI will deploy 6 gigawatts (GW) of AMD GPUs, with the first GW of capacity scheduled to go live in the second half of 2026, using AMD’s upcoming Instinct MI450 series GPUs.
The multi-year, multi-generation partnership is designed to make AMD a core compute partner for OpenAI’s expanding infrastructure needs. It represents one of the largest technology collaborations in the history of artificial intelligence, expected to generate tens of billions of dollars in annual AI revenue for AMD over time while significantly accelerating OpenAI’s infrastructure buildout.
“AMD’s collaboration with OpenAI marks a new era of high-performance, scalable AI compute infrastructure that will redefine what’s possible in generative intelligence.”
The deal highlights AMD’s open and scalable architecture, its leadership in high-performance data center GPUs, and its robust ROCm open software platform. AMD’s rack-scale systems will form the backbone of OpenAI’s compute clusters, optimized for the ever-increasing demand of training and deploying large-scale generative AI models.
To further align strategic interests, AMD has also issued OpenAI a performance-based warrant for up to 160 million shares of AMD stock, which will vest as specific deployment milestones are achieved.
This collaboration underscores AMD’s growing influence in the AI compute ecosystem, positioning it as a defining player in the race to build and power the world’s most advanced AI infrastructure. By combining AMD’s next-generation GPU technology with OpenAI’s pioneering models, the partnership aims to unlock unprecedented performance, efficiency, and scalability—laying the groundwork for the next wave of AI innovation worldwide.
