Alphabet Challenges Nvidia Dominance With Meta TPU Deal
Alphabet is negotiating a multibillion-dollar deal to supply custom AI chips to Meta Platforms, challenging Nvidia's market dominance in the AI hardware sector.
Alphabet Inc. is aggressively challenging the market dominance of Nvidia Corporation by expanding the commercial availability of its custom Tensor Processing Units (TPUs). In a strategic pivot from internal-only use, Alphabet is negotiating a multibillion-dollar agreement with Meta Platforms Inc. to supply TPUs for Meta's data centers starting in 2027, with potential cloud-based leasing as early as 2026. This move aims to provide a cost-effective alternative to Nvidia's GPUs, potentially capturing up to 10% of Nvidia's annual revenue.
Google has further bolstered its position by releasing the Gemini 3 AI model, which was trained entirely on in-house TPUs, and securing a deal to supply up to one million chips to Anthropic PBC. To facilitate adoption, Google enhanced its JAX software and partnered with Broadcom Inc. to develop the seventh-generation Ironwood TPU. These developments contributed to a market rotation where Alphabet's valuation approached $4 trillion, while Nvidia and AMD saw significant stock declines, with Nvidia losing hundreds of billions in market value during intraday trading.
Nvidia has defended its position, asserting that its platform remains "a generation ahead of the industry" and offers superior versatility compared to specialized ASICs. CEO Jensen Huang argued that custom chips struggle with complex inference tasks and noted that Nvidia continues to supply Google. Despite the competition, analysts suggest the shift may be driven by supply constraints and a broader industry trend among hyperscalers—including Amazon and Microsoft—to diversify hardware to reduce vendor lock-in and lower infrastructure costs.