Google and Meta Forge AI Chip Alliance
Google has signed a multibillion-dollar deal to supply Meta with AI chips, creating a powerful new alliance to challenge Nvidia's market dominance. The partnership signals a strategic push by two of the biggest names in tech to gain more control over their AI hardware supply chains and reduce their reliance on a single vendor.
The alliance will see Meta not only rent Google's Tensor Processing Units (TPUs) in a multi-year, multi-billion dollar agreement, but also explore purchasing them for its own data centers starting next year. This move is part of a massive AI infrastructure investment by Meta, with capital expenditures projected to be between $115 billion and $135 billion in 2026, a significant increase from $72 billion in 2025. This partnership follows Meta's reported setbacks in developing its own advanced in-house AI chips. The company has reportedly halted the development of its most advanced training chip, codenamed "Olympus," and altered plans for another chip named "Iris" due to concerns about complexity and the ability to match Nvidia's performance. While Meta continues to develop its Meta Training and Inference Accelerator (MTIA) for specific workloads like recommendation engines, this deal with Google signals a move to secure a reliable supply of high-performance chips for training its large-scale AI models. The deal is a direct challenge to Nvidia, which holds a dominant 92-95% market share in the AI data center chip industry. The mere rumor of this deal in December wiped $250 billion from Nvidia's market value, underscoring the perceived threat to its market position. By diversifying its suppliers, Meta gains leverage in negotiating prices and mitigates risks associated with relying on a single vendor. For Google, this partnership is a significant win for its cloud division and its ambitions to commercialize its custom-designed TPUs. Originally developed for internal use to power services like Search and Photos, Google is now actively leasing TPUs to external companies. Google Cloud executives have set a goal of capturing up to ten percent of Nvidia's annual revenue through TPU sales. Meta's strategy now involves a multi-pronged approach to its AI hardware needs. In addition to the Google deal, the company has also signed significant agreements with Nvidia for its Blackwell and Rubin GPUs and with AMD for up to $60 billion worth of its AI chips. This diversification aims to ensure Meta has the massive computing power required for its ambitious AI and metaverse plans.
