Meta and AMD announced a multiyear strategic partnership to deploy up to 6 gigawatts of AMD Instinct GPUs across Meta's AI data centers — equivalent to powering dozens of hyperscale facilities and one of the largest AI compute commitments in tech history. Bloomberg estimates the deal is worth tens of billions of dollars over at least four years.

Shipments begin in the second half of 2026 using custom AMD Instinct MI450 GPUs and 6th Gen EPYC "Venice" CPUs built on the co-developed Helios rack-scale architecture. As part of the deal, AMD issued Meta a performance-based warrant for up to 160 million AMD shares, vesting as GPU shipment milestones are achieved. AMD stock jumped 11% on the news.

The deal signals Meta's shift toward a multi-supplier strategy: AMD for inference workloads, Nvidia for training, and Meta's own MTIA custom silicon. CEO Mark Zuckerberg called it a "long-term partnership to deploy efficient inference," while AMD CEO Lisa Su highlighted the aligned roadmaps for "high-performance, energy-efficient" AI infrastructure.