OpenAI reportedly plans to buy Google chips to diversify suppliers

OpenAI plans to purchase Google’s tensor processing units, or TPUs, according to a person familiar with the matter cited by The Information. Neither company has confirmed the plan, and pricing, purchase volume and delivery timing remain undisclosed. Owning the processors would help OpenAI reduce its dependence on Nvidia, which supplies most of the hardware supporting the AI boom.

Reports published in June 2025 said OpenAI had begun renting Google’s TPUs, an effort OpenAI described as preliminary testing without plans for large-scale deployment. A San Francisco job advertisement seeking an engineer to optimize frontier models for TPUs supports the possibility of broader adoption. The role includes compiler work, highlighting the software changes required to move models between chip platforms.

OpenAI is also developing Jalapeño, a custom inference chip built with Broadcom, with initial deployment planned by the end of 2026. Its hardware strategy includes arrangements with Nvidia and AMD. Nvidia ultimately invested $30 billion in OpenAI’s funding round, below the up to $100 billion outlined in an earlier letter of intent.

Google is expanding external access to its processors, including systems installed in customers’ own data centers. Alphabet’s CFO expects 2027 to account for most TPU system sales revenue. Rival Anthropic has also agreed with Google and Broadcom to secure multiple gigawatts of next-generation TPU capacity.