OpenAI plans to buy Google's AI chips, The Information reports

What was reported

Reuters reported that The Information said OpenAI plans to buy Google's AI chips, citing a person familiar with the matter. The planned purchase would deepen ties between OpenAI and Alphabet's Google, which compete in AI products and search.

The report did not disclose financial terms or other details. Google did not immediately respond to a Reuters request for comment outside regular business hours, and OpenAI declined to comment.

The Information's reporting on OpenAI and Google has regularly driven AI hardware discussions this year. The outlet previously reported on OpenAI's negotiations with multiple chip vendors and its attempts to secure compute capacity at scale as demand for AI training and inference has exploded. This latest report fits that pattern of an AI lab aggressively expanding its supply options.

Why it matters

OpenAI is one of the largest buyers of Nvidia GPUs, which it uses to train models and run inference. Buying Google's Tensor Processing Units would add a second major chip supplier to its stack, reducing its dependence on Nvidia and giving Google a significant new revenue stream.

Reuters reported in June 2025 that OpenAI had begun renting Google's AI chips, so a move to purchase them would extend an existing relationship from leasing into direct procurement.

The deal would also be notable because OpenAI and Google compete directly on AI products and search, yet the two companies are increasingly intertwined commercially.

The move would also reshape the competitive geography of AI silicon. OpenAI and Google compete head-to-head in AI assistants, search and cloud, yet the two companies have been building increasingly close commercial ties. OpenAI already uses Google Cloud infrastructure, and Google is a major investor in OpenAI's competitors via its own AI ventures. A chip purchase would add a direct financial link on top of those relationships.

The chip context

Google's TPUs have gained traction as an alternative to Nvidia's market-leading AI chips as companies seek cheaper options and chips suited to running AI models. Google began selling TPUs directly to some customers this year, opening a new business beyond its own cloud.

The report arrives as Alphabet prepares to report quarterly results on 28 October, with investors looking for signs that its heavy AI investment is driving cloud growth.

For Nvidia, the report is a reminder that its largest customers are actively exploring alternatives. OpenAI signing up for Google TPUs would be the highest-profile defection from the Nvidia-only path.

The report also fits a wider pattern of AI labs diversifying their chip supply. Amazon, Google, Microsoft and Meta all design their own accelerator chips, and OpenAI itself has been linked to various custom silicon efforts. Adding Google TPUs to the mix would give OpenAI access to Google's massive in-house AI infrastructure expertise and its large TPU fleets.

Analysts have noted that Google's TPU strategy has evolved from being purely an internal advantage for running its own services to a commercial offering. Google began renting TPUs to cloud customers years ago, and this year started selling the hardware directly. A direct sale to OpenAI would be the highest-profile validation of that strategy.

What to watch

Watch Alphabet's earnings on 28 October for any confirmation or commentary on TPU sales to external customers.

Watch whether OpenAI confirms the purchase or discloses terms, which neither company did when the report was published.

Background on OpenAI's compute strategy

OpenAI has been a major consumer of Nvidia hardware since its founding, and its training runs have pushed the limits of available GPU supply. The company has signed large compute deals with Microsoft and Oracle, and it has been reported to be in talks with multiple other providers to secure capacity.

Google TPUs would not replace Nvidia GPUs overnight. Nvidia's CUDA ecosystem and its dominance in training mean most frontier labs still rely heavily on Nvidia silicon. But TPUs have advantages for specific inference workloads and at lower cost, which is why Google has been expanding TPU availability beyond its own cloud.

A purchase agreement would also have strategic value for Google beyond the revenue. Selling TPUs to OpenAI, one of its biggest AI rivals, would legitimize Google's chip lineup as a true alternative to Nvidia, strengthening its position in the AI hardware market and potentially attracting other large customers.

The bigger picture

The AI hardware market is undergoing its biggest shift since Nvidia's rise to dominance. Hyperscalers are designing custom silicon, startups are building accelerators, and even the largest AI labs are considering multi-vendor strategies. If OpenAI adds Google TPUs to its hardware portfolio, it would be the clearest sign yet that no single vendor can count on its biggest customer exclusively.

For investors, the report is another data point in the AI spending supercycle. OpenAI's compute bills have been a major driver of Nvidia's growth, and a shift of even part of that demand to Google would have implications for both companies' earnings. Google reports results on 28 October, and any mention of TPU sales would be closely watched.