Meta has announced a substantial multiyear agreement to purchase up to $100 billion worth of chips from Advanced Micro Devices (AMD), a deal that could necessitate roughly six gigawatts of data center power demand.

This significant partnership underscores Meta's intensified focus on artificial intelligence infrastructure and its strategy to diversify its compute capabilities.

As part of this expansive agreement, AMD has structured a performance-based warrant for Meta, granting it the option to acquire up to 160 million shares of AMD common stock.

This constitutes approximately 10% of thecompany's total shares, each priced at a nominal $0.01.

The vesting of these stock awards is contingent upon specific milestones, and the full issuance is conditional on AMD's share price reaching $600, as reported by The Wall Street Journal.

At the time of the announcement, AMD's stock closed at $196.60, indicating a substantial future appreciation would be required for Meta to realize the full equity benefit.

The chips Meta is slated to acquire under this agreement include AMD’s MI540 series of Graphics Processing Units (GPUs) and its latest generation of Central Processing Units (CPUs).

Notably, CPUs are increasingly recognized as a foundational component of the AI inference compute stack.

Their growing importance stems from their efficiency, ease of scalability, and their role in enabling companies to avoid being exclusively tied to a single vendor like Nvidia, which has historically dominated the AI chip market.

AMD CEO, Lisa Su emphasized the burgeoning demand for CPUs during an investor briefing, stating, “The CPU market is absolutely on fire.”

She highlighted the significant and continuous growth driven by the scaling of AI infrastructure deployments, particularly in inferencing and agentic AI applications, asserting that AMD's portfolio is exceptionally well-positioned to meet these needs.

This surge in demand allows AMD to further gain traction as AI firms actively seek alternatives to reduce their reliance on Nvidia, which has commanded a premium for its leadership in AI chips. AMD had previously secured a similar equity-for-chips deal with OpenAI last October, further solidifying its competitive standing.

Meta CEO, Mark Zuckerberg, described the partnership with AMD as "an important step" in the company's broader objective to diversify its compute resources and advance toward "personal superintelligence."

Zuckerberg defines personal superintelligence as advanced AI systems engineered to profoundly comprehend and empower individuals in their daily lives, signifying a future where AI acts as a deeply integrated personal assistant.

This substantial investment with AMD is part of Meta's larger commitment to developing robust AI infrastructure.

The company has pledged to invest at least $600 billion in U.S. data centers and AI infrastructure over the coming years, with a projected capital expenditure of $135 billion earmarked for 2026 alone.

Illustrating this commitment, Meta recently unveiled plans for a $10 billion gas-powered data center campus in Indiana, designed to support 1 gigawatt of compute capacity.

The AMD partnership follows closely on the heels of another significant multiyear deal Meta struck with Nvidia just weeks prior, aimed at expanding its data centers with millions of Nvidia's latest CPUs and GPUs.

Furthermore, Meta is actively pursuing the development of its own in-house chips, though these efforts have reportedly encountered delays.

These parallel strategies demonstrate Meta's multi-pronged approach to securing the immense compute power required for its ambitious AI goals, leveraging both established and emerging providers while also cultivating proprietary solutions.