Notebookcheck Logo

Meta signs major AI infrastructure deal with AMD

Meta and AMD say the partnership will deploy up to 6GW of AMD Instinct GPUs across Meta’s AI infrastructure, with initial shipments targeted for the second half of 2026.
ⓘ about.fb.com
Meta and AMD say the partnership will deploy up to 6GW of AMD Instinct GPUs across Meta’s AI infrastructure, with initial shipments targeted for the second half of 2026.
Meta and AMD have announced a multi-year AI infrastructure partnership to deploy up to 6GW of AMD Instinct GPUs. AMD says the first 1GW shipments are expected in 2H 2026, and the deal includes a performance-based warrant tied to deliveries.

Meta and AMD say they’ve signed a definitive multi-year, multi-generation partnership to deploy up to 6 gigawatts of AMD Instinct GPUs across Meta’s next-generation AI infrastructure.

Both companies describe the deal as part of Meta’s effort to diversify its AI compute stack and align roadmaps across GPU/CPU silicon, systems, and software.

When hardware shipments start, and what’s in the first deployment

AMD says shipments supporting the first 1-gigawatt deployment are expected to begin in the second half of 2026, using a custom Instinct GPU based on AMD’s MI450 architecture and built on the Helios rack-scale architecture, which the companies previously discussed through the Open Compute Project.

AMD also says the first deployment will pair the MI450-based GPU with 6th-gen EPYC CPUs codenamed “Venice,” running ROCm software.

The warrant clause: Meta could end up with a sizeable stake

AMD’s press release includes a notable financial sweetener: a performance-based warrant for up to 160 million AMD shares, structured to vest as shipment milestones are met (starting with the first 1GW) and as purchases scale toward 6GW, with additional technical and commercial conditions.

Financial Times and AP both report that the structure could translate into as much as a 10% stake in AMD if the warrant is fully exercised.

What each company says it’s trying to achieve

Meta says the partnership is meant to strengthen its AI infrastructure with more flexibility and resilience by using multiple partners, rather than relying on a single vendor.

AMD, for its part, frames the agreement as a long-term ramp that spans GPUs, EPYC CPUs, and rack-scale systems designed around Meta’s workloads.

Why this matters in the broader AI hardware race

The deal lands as large AI buyers continue to lock in multi-year capacity amid persistent demand for accelerators and the build-out of new data centers. Reporting from the Financial Times highlights the scale of Meta’s planned AI infrastructure spending and its push to diversify its supplier base.

Please share our article, every link counts!
Mail Logo
Google Logo Add as a preferred
source on Google
static version load dynamic
Loading Comments
Comment on this article
Darryl Linington, 2026-02-25 (Update: 2026-02-25)