Meta’s $100bn AI chip pact expands computing ambitions

Meta Platforms has struck a landmark long-term agreement with Advanced Micro Devices to buy tens of billions of dollars’ worth of AI processors and data-centre hardware over the next five years, a deal that could exceed $100 billion and reshape the competitive landscape for artificial-intelligence infrastructure. This multiyear pact will see Meta deploy up to six gigawatts of AMD-based computing capacity, starting with initial shipments in the second half of 2026, underlining the company’s strategy to diversify its chip supply and accelerate its AI ambitions.

The agreement moves AMD closer to challenging Nvidia’s longstanding dominance in the AI chip market by positioning its MI450-class GPUs and next-generation CPUs at the heart of one of the world’s largest AI computing fleets. Meta will receive performance-based warrants to acquire up to 160 million shares of AMD common stock at a nominal price, contingent on technical and commercial milestones, potentially giving Meta about a 10 per cent equity stake in the chipmaker. The structure aligns the two companies’ interests around execution and market success.

Meta’s chief executive, Mark Zuckerberg, framed the deal as a pivotal step in building out the company’s infrastructure to support “personal superintelligence” — a vision of deeply integrated, highly capable AI systems running across its platforms. By forging closer ties with AMD, Meta aims to secure a more resilient and flexible compute stack that can accelerate development and deployment of advanced AI workloads while mitigating supply risks associated with relying on a single chip supplier.

The scale of the agreement reflects a broader trend in the technology industry, where hyperscale firms are pushing heavy investments into AI compute capacity to support next-generation services. One gigawatt of compute capacity can draw power comparable to that of a small city, and a six-gigawatt commitment speaks to the immense energy and hardware demands of high-performance AI applications. Companies including Meta, Google, and Microsoft have collectively earmarked hundreds of billions of dollars for data-centre and AI infrastructure buildouts as demand for generative AI and machine learning services grows.

AMD’s chief executive, Dr Lisa Su, said the expanded partnership will include a broader range of AMD technology, including high-performance EPYC CPUs and co-developed rack-scale systems, aligning hardware, software and systems roadmaps to deliver energy-efficient and scalable infrastructure optimised for Meta’s AI workloads. This collaboration extends beyond GPU procurement to encompass end-to-end infrastructure solutions meant to accelerate AI deployment at scale.

Financial markets reacted positively to the announcement, with AMD’s share price jumping by double-digit percentages in early trading, reflecting investor confidence that the deal validates AMD’s role in the high-end AI compute market and provides a stable future revenue stream. Analysts noted that the performance-based warrant structure mitigates some risks for AMD while signalling strong demand from a major technology purchaser.

While Meta has concurrently expanded its ties with Nvidia and continues to develop in-house silicon, the AMD pact is significant for its scale and strategic implications. It broadens Meta’s supplier base at a time when tech firms are eager to secure diverse sources of compute power without being wholly dependent on a single vendor. Nvidia’s GPUs have historically dominated AI training and inference workloads, but AMD’s growing footprint in this space offers customers an alternative amid supply constraints and escalating costs.

Industry watchers have noted that Meta’s infrastructure push includes not only hardware procurement but also extensive investment in new data-centre campuses and partnerships with cloud and hardware providers. The company’s broader AI infrastructure plan, which encompasses a multibillion-dollar capital expenditure programme, aims to support both internal innovations and external services across its social media and metaverse platforms.