Meta Signs $100 Billion Deal With AMD for AI Data Center Chips

Meta has signed a multi-year agreement with AMD worth up to $100 billion for AI data center processors, in one of the largest semiconductor deals in history. The agreement, announced this week, would see Meta acquire six gigawatts worth of AMD chips to power its sprawling AI infrastructure — and includes a performance-based warrant granting Meta up to 160 million AMD shares (roughly 10 percent of the company) at $0.01 per share, with vesting tied to deployment milestones.

A Seismic Shift in the AI Chip Market

The deal comes just days after Meta purchased millions of Nvidia's latest AI chips, signaling that Mark Zuckerberg's company is pursuing an aggressive dual-supplier strategy for its artificial intelligence ambitions. Rather than relying solely on Nvidia — which has long dominated the AI accelerator market — Meta is diversifying its supply chain with a massive bet on AMD.

This mirrors a broader trend across the tech industry. AMD previously signed a similar multi-year agreement with OpenAI, positioning itself as the primary alternative to Nvidia in the high-performance AI compute space. With both Meta and OpenAI now committed to AMD hardware at scale, the chipmaker is rapidly closing the gap with its larger rival.

What Six Gigawatts Means

To put the scale of this deal in perspective, six gigawatts of computing power is enough to run multiple hyperscale data centers simultaneously. Meta has been on a data center building spree, expanding its infrastructure across the United States and internationally to support its AI models, including its Llama family of open-source large language models.

The company's AI investments have accelerated dramatically. Meta CEO Mark Zuckerberg has repeatedly stated that AI is the company's top priority, with billions being poured into training and inference infrastructure. This AMD deal ensures Meta will have the silicon it needs for years to come.

The Equity Component

Perhaps the most striking element of the agreement is the equity stake: the deal includes a performance-based warrant for Meta to acquire up to 160 million shares of AMD common stock — roughly 10 percent of the company — at an exercise price of just $0.01 per share, with vesting tied to deployment milestones. This kind of strategic investment goes beyond a simple procurement contract — it creates a deep financial alignment between the two companies. If AMD's stock rises on the back of AI demand, Meta benefits directly.

For AMD, securing a customer of Meta's scale provides revenue visibility and validates its AI chip roadmap. The company has been investing heavily in its Instinct series of data center GPUs and its EPYC server CPUs, both of which are likely central to this agreement.

Breaking Nvidia's Grip

Nvidia has enjoyed near-monopoly status in the AI training chip market, with its H100 and subsequent Blackwell-generation GPUs becoming the gold standard for AI workloads. However, the sheer demand for AI compute — and the supply constraints that come with it — have pushed major tech companies to seek alternatives.

Google has its custom TPU chips. Amazon has Trainium and Inferentia. Microsoft has its Maia AI accelerator. Now Meta, rather than building custom silicon, has chosen to invest massively in AMD as its second pillar alongside Nvidia.

What's Next

The deal is expected to unfold over multiple years, with deliveries ramping up as AMD scales production. Industry analysts expect this agreement to put significant pressure on Nvidia's pricing power and may encourage other hyperscalers to negotiate similar large-scale AMD commitments.

For the AI industry as a whole, greater competition in the chip market is welcome news. More supply options mean lower costs, faster innovation, and fewer bottlenecks in the race to build increasingly powerful AI systems.

The Meta-AMD deal is a clear signal: the AI infrastructure buildout is far from over, and the companies willing to write the biggest checks will shape the future of computing.