Google & Meta Team Up on AI Chips
Google has signed a multibillion-dollar deal to supply Meta with advanced AI chips. The partnership is a direct challenge to Nvidia's market dominance and aims to accelerate the deployment of next-gen AI models across Meta's apps.
This partnership arrives as Meta shelves its most advanced in-house AI chip projects. The company has reportedly faced significant technical challenges and delays with its "Olympus" and "Iris" chips, making it more reliant on external suppliers to power its AI ambitions. The deal centers on Google's custom-designed Tensor Processing Units (TPUs), an alternative to the GPUs that dominate the market. Initially developed for Google's internal services like Search and Gemini, TPUs are specialized for AI workloads and are now being leased to external companies, positioning Google Cloud as a key player in the AI infrastructure race. This move is part of a broader diversification strategy for Meta, which is also making massive investments with other chipmakers. The company recently announced a multi-billion dollar deal with AMD for its Instinct AI chips and continues to be a major customer for Nvidia's next-generation processors. Nvidia currently holds a commanding position in the AI chip market, with a market share estimated to be between 80% and 85%. This dominance has created a highly competitive environment, with tech giants like Meta seeking multiple partners to mitigate supply chain risks and secure the massive computing power needed for future AI development. Under the terms of the multi-year agreement, Meta will rent the TPUs, but the two companies are also reportedly in discussions for Meta to purchase the chips directly for its own data centers as early as 2027. This would mark a significant shift, as Google has historically reserved its custom silicon for its own use or for cloud-based rentals.
