Pranav Hotkar
Menlo Park, United States - February 24, 2026 - Meta Platforms has entered a multi-year strategic partnership with Advanced Micro Devices to deploy up to 6 gigawatts of AI compute infrastructure, marking one of the largest silicon-backed data center buildouts announced to date.
The collaboration will combine AMD Instinct GPUs, EPYC CPUs and rack-scale systems tailored for Meta’s artificial-intelligence training and inference workloads. Initial deployments totalling about 1 gigawatt are expected to begin shipping in the second half of 2026, with capacity expanding in phases over multiple hardware generations.
Meta CEO Mark Zuckerberg said the agreement supports the company’s effort to broaden its infrastructure stack. “An important step for Meta as we diversify our compute,” he said in the announcement.
AMD chair and CEO Lisa Su described the partnership as a large-scale deployment aligned across hardware and software roadmaps. “Accelerating one of the industry’s largest AI deployments,” she said.
The systems will integrate AMD’s upcoming Instinct accelerator architecture with next-generation EPYC processors and an open rack-scale platform co-developed through industry standards initiatives. The deployment will also rely on AMD’s ROCm software ecosystem to support model training and inference at hyperscale.
The companies indicated the agreement could extend across multiple data center campuses and is tied to long-term capacity milestones. Industry analysts say the scale, measured in gigawatts rather than server counts, reflects how AI infrastructure planning is increasingly linked to power availability and facility build-out rather than only chip shipments.
The partnership also signals a broader shift among hyperscale operators toward diversifying silicon supply chains as AI demand accelerates. By coordinating system design directly with a chipmaker, operators aim to optimise performance, deployment timelines, and energy efficiency across entire campuses instead of individual hardware layers.
Meta has been rapidly expanding its AI infrastructure footprint to support large-model training and production inference services across its platforms. The new agreement positions AMD as a primary supplier for a significant share of that expansion while reinforcing the trend toward vertically coordinated compute architectures spanning silicon, software, and facilities.
