AMD and Meta have entered into a long term agreement, AMD will provide Meta with 6GW(roughly enough to power 5 million houses) of AMD Instinct GPUs. The move is part of Meta’s effort to build a flexible tech stack with multiple partners to run different workloads.

This scale is intended to to support Meta’s ambitious goal of delivering personal superintelligence to its billions of users. The deployment is scheduled to begin in the second half of 2026.

This isn’t just a typical hardware purchase, the two companies are aligning their long term roadmaps to across 3 critical areas. Silicon computing – Utilizing AMD Instinct GPUs and EPYC CPUs. Systems – Using the Helios rack scale architecture that was co-developed with AMD. Software – Ensuring vertical integration so Meta’s software stack runs natively and efficiently on AMD’s hardware.

By bringing AMD into the picture, Meta is reducing its reliance on a sole chip supplier (traditionally NVDIA). The strategy involves a mix of third party chips(AMD’s Instinct line) and Meta’s custom in-house silicon chips(Training and Inference Accelerator (MTIA) silicon).

For Meta, this is about future proofing their infrastructure to handle heavy AI workloads. For AMD, it cements their position as a primary competitor in the high tech AI chip market.