Meta’s $100B AMD Deal Powers AI ‘Superintelligence’ Push
▼ Summary
– Meta plans to purchase up to $100 billion worth of AMD chips, a deal that would drive about six gigawatts of data center power demand.
– As part of the agreement, AMD issued Meta a warrant for up to 160 million shares, which vest based on performance milestones and AMD’s stock price reaching $600.
– Meta will purchase AMD’s MI540 GPUs and latest CPUs, with CPUs becoming a core AI inference pillar due to their efficiency and scalability.
– This partnership is part of Meta’s strategy to diversify its compute infrastructure away from reliance on Nvidia and work toward developing “personal superintelligence.”
– The deal follows Meta’s recent multiyear agreement with Nvidia and its broader pledge to invest hundreds of billions in U.S. data centers and AI infrastructure.
In a landmark move for the artificial intelligence industry, Meta has entered into a multiyear agreement to purchase potentially up to $100 billion worth of AMD processors. This colossal deal is expected to fuel approximately six gigawatts of data center power demand, underscoring the immense energy requirements of next-generation AI infrastructure. The partnership represents a strategic shift as major tech companies actively seek alternatives to reduce their dependence on a single chip supplier.
As part of the arrangement, AMD has granted Meta a performance-based warrant for up to 160 million shares of its common stock, priced at just one cent each. This equity award, constituting about ten percent of AMD, is structured to vest as Meta meets specific purchasing milestones. A notable condition requires AMD’s share price to reach $600 before Meta can receive the final tranche of stock, a significant jump from its recent closing price near $196. Meta’s procurement will focus on AMD’s MI540 series of graphics processing units (GPUs) alongside its newest central processing units (CPUs).
AMD CEO Lisa Su emphasized the surging demand for CPUs, describing the market as “absolutely on fire.” She attributed this growth directly to the scaling of AI infrastructure, particularly for inference workloads and the development of agentic AI systems. CPUs are gaining prominence in AI compute stacks due to their efficiency, scalability, and the flexibility they offer companies not wanting to be locked into a single vendor’s ecosystem. This deal follows a similar pattern from last October, when AMD and OpenAI exchanged equity for a chip supply agreement, highlighting a broader industry trend.
For Meta, this partnership is a cornerstone of its ambitious AI roadmap. CEO Mark Zuckerberg called it “an important step” in diversifying the company’s compute resources as it works toward developing what he terms “personal superintelligence.” Zuckerberg envisions these as AI systems profoundly tailored to understanding and assisting individuals in their daily activities. The commitment is backed by Meta’s pledge to invest at least $600 billion in U.S. data centers and AI infrastructure in the coming years, with a projected capital expenditure of $135 billion in 2026 alone. Recent projects include a planned $10 billion, one-gigawatt data center campus in Indiana.
This strategic alignment with AMD arrives just weeks after Meta finalized another major multiyear deal to acquire millions of the latest CPUs and GPUs from Nvidia, the current dominant force in AI chips. The dual-supplier strategy highlights Meta’s urgent drive to secure vast and diverse computing power while it continues developing its own in-house chip designs, a project that has reportedly encountered delays. Together, these investments paint a picture of a company aggressively allocating resources to lead the coming wave of advanced artificial intelligence.
(Source: TechCrunch)
