The Information and Bloomberg report that OpenAI plans to buy Google's AI chips
The reports say financial terms for the potential deal are unavailable
Reuters reported last year that OpenAI had previously rented Google's TPUs
Google will bring its TPUs to Oracle Cloud, with deployments starting in 2027
OpenAI has built most of its infrastructure on Nvidia GPUs
OpenAI plans to purchase Google's AI chips, according to The Information and Bloomberg, as the ChatGPT maker seeks to expand its computing capacity.
The financial terms of the potential deal are not yet available, the reports said, citing sources familiar with the matter.
The AI lab has previously rented Google's tensor processing units (TPUs), which were historically developed for internal use, Reuters reported last year.
The potential chip purchase comes as Google's TPUs have increasingly emerged as an alternative to Nvidia's chips. While Google initially developed the chips primarily for internal use, the tech giant has recently begun expanding their availability to external customers.
On Wednesday, Google announced a partnership with cloud computing firm Oracle to bring its TPUs to Oracle Cloud Infrastructure, with deployments starting in 2027.
"As AI workloads scale across every industry, organizations need compute options that optimize throughput, latency, and unit economics across the AI lifecycle," said Google.
Google's TPU customers include Anthropic, which has made multi-year commitments involving the chips, and Meta, which has reportedly signed a multi-billion-dollar deal to rent TPUs.
Meanwhile, OpenAI remains one of Nvidia's major customers and has built most of its infrastructure on Nvidia GPUs. In August, Nvidia reportedly agreed to provide a guarantee of up to $105 billion to help OpenAI lease a data center in Ohio.
The Latent has reached out to OpenAI and Google for comment.
