By Hamza Faisal
OpenAI plans to buy Google's AI chips, The Information reported on Tuesday, in a move that would deepen one of the most surprising partnerships in the artificial intelligence industry — and loosen Nvidia's grip as OpenAI's most important silicon supplier.
The planned purchase, reported by The Information citing a person familiar with the matter, would see the ChatGPT maker buy Google's Tensor Processing Units, or TPUs, outright rather than simply renting computing power through Google Cloud. Financial terms and other details of the planned purchase could not be learned, and neither company would discuss it: OpenAI declined to comment, and Google did not immediately respond to a request for comment.
OpenAI is one of the largest buyers of Nvidia's graphics processing units, the chips used to train AI models and serve hundreds of millions of users. That dependence is exactly what makes the Google move striking: the two companies compete directly in AI products and search, yet OpenAI is now preparing to build part of its future infrastructure on its rival's silicon.
From renting to buying
The shift escalates a relationship that only recently became public. Reuters reported in June 2025 that OpenAI had begun renting Google's AI chips through Google Cloud to run ChatGPT and other products, its first meaningful use of non-Nvidia hardware, with the goal of reducing inference costs. At the time, Google was not renting its most powerful TPUs to OpenAI.
Buying the chips outright is a different kind of commitment. It would give OpenAI direct control over a slice of its compute capacity rather than a short-term patch for peak demand. Google, for its part, began selling TPUs directly to some customers this year, turning a technology long kept in-house into a product line in its own right.
The clearest public signal of OpenAI's intent is a job listing. The company is hiring a Software Engineer, TPU Compilers, in San Francisco, with the role described as leading the optimization of its frontier models on TPUs.
Why TPUs, why now
For OpenAI, the logic is diversification. TPUs have gained traction as an alternative to Nvidia's market-leading chips as companies seek cheaper options and hardware better suited to running models at scale. Inference — the everyday computing used to answer user prompts — is where efficiency matters as much as raw speed, and it is where OpenAI's costs are most sensitive to volume.
The hiring pattern points the same way. OpenAI appointed Richard Ho, who was senior director of engineering at Google and led development of its Cloud TPUs, to run its hardware division. A compiler team working on TPU optimization would presumably sit inside that organization, building the software bridge between OpenAI's models and Google's silicon.
What it means for the chip race
For Google, an OpenAI purchase would be powerful validation that its in-house chips can run other companies' most demanding workloads — not just its own. Alphabet reports its quarterly results later on Wednesday, and evidence that TPUs are becoming a real product rather than an internal tool will be part of that story.
For Nvidia, the message is subtler but unmistakable: its biggest customer is actively building an alternative supply chain. OpenAI is not walking away from Nvidia's GPUs, which remain central to training its models. But the company that once relied on a single silicon supplier is now renting, hiring for, and preparing to buy a rival's chips — and in the AI infrastructure race, that kind of diversification tends to compound.



