Meta Signs Multi-Year AMD Deal for 6 Gigawatts of AI Compute

Meta

Feb 24, 2026 · Updated Apr 25, 2026

Meta signed a multi-year deal with AMD to deploy up to 6 gigawatts of Instinct GPUs across its AI infrastructure, with shipments starting H2 2026. The agreement signals Meta's push to diversify its compute stack beyond a single GPU supplier.

Meta signed a multi-year, multi-generation agreement with AMD to deploy up to 6 gigawatts of AMD Instinct GPUs, EPYC CPUs, and rack-scale AI systems across its global data center network. The first deployments use the Helios rack-scale architecture that Meta and AMD co-developed at the Open Compute Project summit, with shipments beginning in H2 2026.

This is part of Meta's portfolio-based infrastructure strategy - diversifying GPU suppliers instead of relying on a single vendor. Meta is combining AMD hardware with a range of other hardware partners and its own Meta Training and Inference Accelerator (MTIA) silicon program to build a flexible, resilient compute stack.

AMD called this one of the industry's largest AI deployments. The deal extends to roadmap alignment across silicon, systems, and software — meaning AMD and Meta will co-develop future hardware generations together.

View the full update on about.fb.com

AI at Meta

@AIatMetaFeb 24

Meta 🤝 AMD Today we’re announcing a multi-year agreement with @AMD to integrate their latest Instinct GPUs into our global infrastructure. With approximately 6GW of planned data center capacity dedicated to this deployment, we’re scaling our compute capacity to accelerate the development of cutting-edge AI models and deliver personal superintelligence to billions around the world. Learn more: https://t.co/GFggxfNZ8w

317

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from Meta →

Keep reading

Meta Releases MTIA Chip Roadmap With Four Generations in Two Years

Meta has detailed its MTIA chip family, built with Broadcom, shipping four generations in two years. MTIA 450 doubles HBM bandwidth for GenAI inference and MTIA 500 adds 50% more — both scheduled for mass deployment in 2027.

Zyphra Launches AMD-First Inference Cloud Optimized for Long-Horizon Agents

ZyphraMay 15

Zyphra Launches AMD-First Inference Cloud Optimized for Long-Horizon Agents

Zyphra launched Zyphra Cloud, a full-stack AI platform on AMD MI355X GPUs rather than NVIDIA, opening with serverless inference for long-horizon agents. The 288GB of memory per AMD chip — versus 192GB on NVIDIA's B200 — keeps nearly double the agent sessions resident in VRAM at long context.

Anthropic Secures 5 Gigawatts of Amazon Compute Amid 30 Billion Revenue Surge

AnthropicApr 20

Anthropic Secures 5 Gigawatts of Amazon Compute Amid 30 Billion Revenue Surge

Anthropic expanded its Amazon partnership to secure up to 5 gigawatts of compute capacity and committed $100 billion to AWS infrastructure over the next decade. This massive scale-up follows a revenue surge to $30 billion, providing the physical power needed to stabilize performance for its rapidly growing user base.

NVIDIA Turns AI Factories Into Flexible Grid Assets

NVIDIAMar 23

NVIDIA Turns AI Factories Into Flexible Grid Assets

NVIDIA and Emerald AI are partnering with major energy companies to turn AI data centers into grid-stabilizing assets, potentially freeing up 100 GW of capacity on the existing U.S. power infrastructure.