Meta inks $60B AMD AI pact; 6GW GPUs planned

• Lyla Velez
• February 24, 2026
• News
• 0 Comments

Key Points:

• Meta and AMD sign multiyear AI infrastructure partnership exceeding $100 billion.
• Deal diversifies supply beyond one vendor, accelerates AI data-center compute buildout.
• Contracted sales up to $60B over five years; planning and future capabilities.

Meta Platforms (META) and Advanced Micro Devices (AMD) agreed to a multiyear AI infrastructure partnership valued at more than $100 billion, as reported by The Wall Street Journal. The scale signals an effort to broaden supply beyond a single vendor while accelerating buildout of data-center compute for AI.

Within that broader scope, the contracted sales component totals up to $60 billion over five years, according to Reuters. The difference between the contracted value and the larger headline reflects longer-term infrastructure planning, potential equity-linked features, and future capabilities to be deployed over time.

Scope and mechanics of the Meta AMD AI chips deal

The agreement includes deploying up to 6 gigawatts of GPUs into Meta’s AI data centers, according to CNBC. Multi-gigawatt footprints typically imply staged rollouts across multiple campuses, with networking, storage, and software stacks scaled alongside accelerator deliveries.

Deal mechanics reportedly include stock warrants that could allow Meta to obtain as much as a 10% stake in AMD, contingent on performance and share‑price milestones that include thresholds as high as $600 per share, based on Investing.com. Warrant structures can align incentives by tying potential equity to delivery and adoption outcomes, while also introducing dilution risk if exercised at scale.

AMD has described a roadmap spanning multiple CPU generations alongside custom GPUs tuned for AI inference workloads, indicating a mix of training and inference capability as infrastructure matures. “Meta is taking a big bet on AMD,” said Lisa Su, CEO, outlining the strategic depth of the collaboration.

Energy, supply chain, and delivery considerations

A plan measured in gigawatts concentrates execution risk in power and real estate. Multi‑GW deployments require long‑lead tasks such as grid interconnections, substation buildouts, and contracting firm renewable or capacity arrangements to keep utilization and costs predictable.

Supply chain and product timing are additional variables. Packaging yields, memory availability, and network fabric readiness can affect delivery schedules and performance per watt, which in turn influence total cost of ownership and rollout pacing.

At the time of this writing, based on data from Simply Wall St, AMD’s shares had declined about 3.5% over seven days and 22.9% over 30 days, with a one‑year return of 80.6% and a three‑year return of 156.3%. That background context does not reflect forward guidance and may not capture near‑term market reactions to this agreement.