Commitment to up to 6 gigawatts (6GW) of AMD Instinct GPUs across multiple generations to power Meta’s AI infrastructure (training and especially inference for large models like Llama and future projects). Multi-year, multi-generation agreement: hardware, systems, and software roadmaps are closely aligned between the two companies. First gigawatt: shipments to support the initial deployment are scheduled to begin in the second half of 2026, powered by custom AMD Instinct GPUs based on the MI450 architecture (optimized specifically for Meta’s workloads), integrated into the Helios rack-scale architecture (co-developed with Meta via the Open Compute Project). Meta will also deploy millions of AMD EPYC processors (including the 6th Gen “Venice” CPUs) across its data centers — AMD already powers the majority of Meta’s core services. Creative deal structure: Meta receives warrants (options to purchase) for up to 160 million AMD shares (roughly 10% of the company) at a nominal exercise price of $0.01 per share. These vest progressively based on purchase volumes and certain milestones (first tranche at 1GW shipped, with additional tranches up to 6GW). This aligns Meta even more closely with AMD’s long-term success.
Estimated value: Analysts are quoting a deal worth tens of billions of dollars — with some estimates reaching $60-100 billion over 5+ years, depending on per-watt pricing and final volumes. It’s one of the largest AI infrastructure commitments ever made to a non-Nvidia supplier. Explosive context: This comes just days after Meta announced a major expansion of its partnership with Nvidia (millions of Blackwell GPUs and beyond). Meta is aggressively diversifying its suppliers to avoid single-vendor dependency in the AI race. AMD is gaining serious traction in inference workloads (where power efficiency and cost-per-token matter most) and proving that the MI300/MI400/MI450 series are highly competitive. This is AMD’s second mega-deal of this scale in recent months (following a similar 6GW commitment with OpenAI in late 2025). Lisa Su’s team is executing a clear strategy: deep hyperscaler partnerships + equity alignment to accelerate AI market share gains. Market reaction: AMD stock surged +10-14% in pre-market and early trading (from ~$196 to over $220 in initial moves). Wall Street is clearly rewarding AMD’s growing credibility as a serious AI contender against Nvidia. Why this matters:
For AMD: Massive validation from one of the world’s largest AI spenders → accelerated AI revenue, better order visibility, and a powerful strategic ally. For Meta: Supplier diversification, better pricing/power efficiency on inference, and equity upside in AMD to influence future roadmaps. For the industry: Nvidia’s dominance is being challenged, GPU pricing could stabilize, and innovation accelerates.
AMD is no longer the quiet challenger — it’s becoming a core pillar of the AI revolution. What do you think? Is this the start of a real, credible alternative to Nvidia, or just smart diversification? Official sources: