OpenAI plans to buy Google AI chips in reported supply deal

What was reported, and what was not

The technology news outlet The Information reported on October 27, citing people familiar with the matter, that OpenAI plans to buy Google's artificial intelligence chips, according to RTI's account of the report. The report said the purchase would deepen OpenAI's commercial relationship with Google's parent company Alphabet, even though the two companies compete in AI products and search.

The report did not disclose the financial terms of the transaction or other details. Reuters asked Google for comment outside business hours and did not immediately receive a response, and OpenAI declined to comment, according to the same account. China City Network also reported that OpenAI plans to buy Google AI chips and that the financial terms and other details are unclear.

That is the whole of the confirmed record: a report attributed to unnamed people, reproduced by two outlets, with no company confirmation. A reported plan is not a signed contract, and none of the sources reviewed here gives the size, timing or composition of any purchase.

What Google's chips are, and why they matter

Google's AI chips are tensor processing units, or TPUs, which the company designs for machine-learning workloads. According to RTI's report, TPUs have been gaining favour as companies look for lower-cost alternatives to Nvidia's graphics processors, which have led the market for AI chips. Google began selling TPUs directly to some customers this year.

OpenAI is already one of the largest buyers of Nvidia's GPUs, which are used to train and deploy its models. Reuters reported in June 2025 that OpenAI had begun renting Google's AI chips, so a purchase would extend an existing technical relationship rather than create a new one. The distinction matters: renting capacity is a way to run workloads now, while buying chips is a capital decision that shapes where a company's compute sits for years.

Why a second supplier is a strategic question

For a company of OpenAI's size, the question is not whether it needs more compute but where that compute comes from. Nvidia's GPUs have been the default for training frontier models, and demand for them has outpaced supply through the AI boom. Adding another supplier gives a buyer more ways to secure capacity, more room to negotiate on price, and a fallback if one supply chain slips.

The reported move also fits a broader pattern of AI companies spreading purchases across vendors. RTI notes that Amazon, Microsoft and Alphabet plan to spend up to $600 billion combined this year on AI infrastructure, as they build data-centre capacity for themselves and their customers. Those three companies are simultaneously OpenAI's partners, competitors and potential suppliers, which is why chip decisions carry commercial weight beyond the hardware.

The report frames the purchase as deepening OpenAI's commercial relationship with Alphabet, and that is the only rationale the sources state. It does not describe OpenAI's supplier strategy in the company's own words, so the cost and capacity arguments above are context rather than stated motives.

The spending backdrop

The report lands in a week of heavy AI spending news. According to RTI, Amazon, Microsoft and Alphabet plan to invest up to $600 billion combined this year in AI infrastructure, while investors are increasingly asking when that spending will translate into returns. Alphabet was due to report quarterly results on the evening of October 28, and investors were looking for signs that its large AI investments were continuing to drive cloud growth.

That timing matters for how the chip report should be read. A company buying another company's chips is a customer signal about demand, and Alphabet's earnings were the next opportunity for management to describe that demand. The chip report itself, however, came from unnamed sources and carried no numbers, so it cannot be used to estimate revenue for either company or to forecast how much capacity OpenAI will add.

For Alphabet, the question investors were asking was whether AI spending was still translating into cloud growth, and the chip report arrived one day before that test.

What remains unclear

Several questions are open. The report does not say how many chips OpenAI would buy, what they would cost, when they would be delivered, or whether the arrangement covers training, inference or both. It does not say whether the deal has been signed or is still under discussion, and it does not name the people familiar with the matter.

There is also the question of confirmation. Neither Google nor OpenAI confirmed the purchase in the sources reviewed here, and Reuters' request for comment to Google did not produce an immediate response. Bloomberg's original report was not accessible for this article and is therefore not cited as a source; the account here rests on the RTI report and the China City Network summary, both of which describe the same underlying report.

Finally, a chip purchase is not the same as a chip supply agreement for future years. A single order would add capacity; a long-term agreement would change the supplier mix. The reports describe the former without ruling out the latter.

The report also does not say whether these chips would replace Nvidia orders or add to them, which is the difference between diversifying suppliers and substituting one for another.

How to read the report

The useful facts are narrow and specific: The Information reported a planned purchase on October 27, citing unnamed people; two outlets repeated the report; neither company confirmed it; the terms are undisclosed; and Google has been selling TPUs directly while OpenAI has rented Google chips since at least 2025. Everything else is context.

For readers tracking the AI compute market, the signal to watch is confirmation. A Google or OpenAI statement, a filing, or details in Alphabet's earnings call would turn a reported plan into a disclosed transaction. Until then, the accurate sentence is that OpenAI is reported to plan a purchase of Google AI chips, not that it has bought them.

The broader picture is a market where the largest AI companies are trying to secure compute from multiple suppliers while spending record sums on infrastructure. A report that one of them is talking to a second chip supplier fits that picture. It does not, on its own, show that Nvidia's position has weakened or that Google has won a major customer. Related coverage includes Nvidia antitrust questions, OpenAI models on Google Cloud, and Industry. The site also covered Microsoft's Azure Maia expansion.