Meta and AMD Strike Multi-Year 6GW AI Chip Deal
Meta Platforms and Advanced Micro Devices have signed a multi-year agreement to supply up to 6 gigawatts of AMD Instinct GPUs to power Meta’s artificial intelligence infrastructure. The deal could be worth more than $100 billion, making it one of the largest AI hardware commitments in the industry’s history.
The first deployment under the agreement will use a custom AMD Instinct GPU built on the MI450 architecture, paired with 6th Generation AMD EPYC processors codenamed “Venice.” Shipments supporting the initial one-gigawatt deployment are scheduled to begin in the second half of 2026. The infrastructure will run on AMD’s ROCm software stack and be built on the AMD Helios rack-scale architecture, a platform co-developed by AMD and Meta through the Open Compute Project to support scalable, rack-level AI systems.
Beyond MI450-based GPUs, Meta will also serve as a lead customer for AMD’s next-generation EPYC processor codenamed “Verano,” which is being designed with workload-specific optimizations aimed at delivering strong performance per dollar per watt. Meta has been an AMD customer across multiple generations, having already deployed millions of EPYC CPUs and significant numbers of Instinct MI300 and MI350 series GPUs across its global data center infrastructure.
As part of the financial structure, AMD has issued Meta a performance-based warrant granting the right to purchase up to 160 million shares of AMD common stock at $0.01 per share, representing a potential stake of approximately 10% in AMD. The warrants vest in tranches tied to shipment milestones, with the first tranche unlocking after the initial one-gigawatt delivery and subsequent tranches activating as purchases scale toward the full six-gigawatt target. Vesting is also contingent on AMD reaching defined stock price thresholds and Meta achieving specific technical and commercial milestones. Following the announcement, AMD shares rose more than 9% in pre-market trading.
The AMD agreement is part of Meta’s broader Meta Compute initiative, a strategy to expand its AI infrastructure for what the company describes as the era of personal superintelligence. Rather than relying on a single chip supplier, Meta is pursuing a portfolio approach that combines hardware from multiple external partners alongside its own internally developed silicon, the Meta Training and Inference Accelerator, known as MTIA.
The AMD deal was announced just days after Meta revealed a separate long-term partnership with Nvidia for AI data center chips, reinforcing the scale of Meta’s hardware diversification effort. Meta has been accelerating its AI investments broadly. In mid-2025, the company made a $14.3 billion investment in AI data company Scale AI and recruited its CEO Alexandr Wang to lead a superintelligence team. In December 2025, Meta acquired AI startup Manus as part of its continuing push to expand AI capabilities across Facebook, Instagram, and WhatsApp.
AMD has framed the multi-year agreement as a driver of substantial revenue growth and expects it to be accretive to its non-GAAP earnings per share. The performance-based warrant structure is designed to align both companies’ financial interests around the successful execution of the deployment roadmap.
