Meta Taps AMD for 6GW AI Buildout

In a significant expansion of their strategic collaboration, AMD and Meta have announced a multi-year, multi-generation partnership to deploy up to six gigawatts of AMD Instinct GPUs for Meta's AI infrastructure. This substantial AMD Meta AI partnership aims to accelerate the development and deployment of Meta's cutting-edge AI models, with roadmaps aligned across silicon, systems, and software.

The first phase of this deployment, expected to commence in the second half of 2026, will utilize a custom AMD Instinct GPU based on the MI450 architecture. These GPUs, optimized for Meta's specific workloads, will be paired with 6th Gen AMD EPYC "Venice" CPUs, running on the ROCm software stack and integrated into the AMD Helios rack-scale architecture developed jointly through the Open Compute Project.

Deeper Hardware Alignment

Beyond GPUs, the collaboration extends to AMD's EPYC processors. Meta, a long-standing AMD EPYC customer, will serve as a lead partner for both the 6th Gen "Venice" and the upcoming "Verano" EPYC CPUs. This expanded CPU partnership underscores the strategic role of CPUs in managing the efficiency, scalability, and orchestration of complex AI compute stacks.

To solidify this long-term alignment, AMD has issued Meta a performance-based warrant for up to 160 million shares of AMD common stock. This warrant vests as specific Instinct GPU shipment milestones are met, with the first tranche tied to the initial one-gigawatt deployment. Further vesting is linked to AMD's stock price thresholds and Meta's technical and commercial achievements, directly tying financial incentives to the success of the partnership.

This comprehensive agreement positions AMD at the core of Meta’s ambitious AI buildout, aiming to deliver personal superintelligence and AI-powered services to billions of users globally.