Key takeaways
- The Information reported on October 27 that OpenAI plans to buy Google's TPU chips, and neither company has confirmed the deal or its size.
- OpenAI is hiring engineers to optimize its frontier models for TPUs, though job postings suggest that work is at an early stage.
- Google began selling TPU hardware directly this year and named Oracle as the first outside cloud to offer the chips, available by 2027.
In this story 4 sections
Why would OpenAI, one of Nvidia's biggest customers, want chips from a company it competes with every day? According to The Information, it does. The outlet reported on Tuesday, October 27, that OpenAI plans to buy AI chips from Google, citing a person familiar with the matter. The OpenAI Google TPU plan has no announced size, price or start date, and neither company has confirmed it. OpenAI declined to comment to Reuters, which followed up on the report, and Google did not immediately respond to The Next Web.
Still, the report fits a pattern that has been building for more than a year. Here is how the two rivals got to this point.
From renting Google chips to buying them
- June 2025: Reuters reported that OpenAI had started renting Google's tensor processing units through Google Cloud. OpenAI later said it had no active plans to use the chips in its products, which cooled talk of a shift away from Nvidia.
- April 2026: Google said it would sell TPU hardware directly to a small group of customers, a break from its long habit of offering the chips only as a rented cloud service.
- October 27, 2026: Google announced that Oracle will be the first major cloud company outside Google to offer TPUs to its customers. Oracle said the chips will be available by 2027, according to The Information.
- October 27, 2026: The Information reported OpenAI's plan to buy Google chips. According to Korean outlet DigitalToday, which summarized the report, OpenAI is also hiring engineers to optimize its frontier models for TPUs, and the job postings suggest that work is at an early stage.
Each step lowered a barrier. First OpenAI tried the chips as a renter. Then Google turned the hardware into something a customer could own. Then a rival cloud got access. A direct purchase by the best known AI lab would be the next logical move.
Why a second chip supplier matters to OpenAI
The Next Web described the reported plan as buying Google chips "alongside" Nvidia, not instead of it. That framing matters. Training and running large models takes enormous amounts of compute, and leaning on a single supplier leaves a lab exposed to shortages and pricing it cannot control.
Google competes with OpenAI directly through Gemini, so the arrangement would be unusual. But the AI business already runs on deals between rivals. OpenAI made its models available on Google Cloud earlier this month, as we covered in our report on OpenAI models arriving in Google's Model Garden. Buying hardware would go a step further, because it ties OpenAI's software work to Google's chip design.
The hiring is the detail to watch. Models tuned for Nvidia GPUs do not move to TPUs for free. Code, compilers and performance tuning all need work, which is why job postings for TPU optimization say more about intent than any unnamed source.
What Google gains from selling TPUs
For Google, an OpenAI order would be a loud signal to other buyers. Alphabet CFO Anat Ashkenazi has said she expects only a small share of TPU sales revenue this year, with most of it arriving in 2027, according to The Next Web.
Demand is not the problem. Reuters reported that Amazon, Microsoft and Alphabet plan to spend up to $600 billion combined on AI infrastructure this year. Amazon has said its own capacity is largely spoken for, as we noted in our look at AWS capacity through 2027. In a market that tight, a second source of high end chips is valuable to nearly everyone except, perhaps, Nvidia.
Three signs that would confirm the OpenAI Google TPU deal
- A statement from OpenAI or Google naming the chip purchase.
- Details on volume, which would show whether this is a test order or a major commitment.
- Evidence that OpenAI models are being served on TPUs in production.
There is also the money question. OpenAI's rivals are raising huge sums to pay for compute, as seen in the numbers inside Anthropic's IPO filing and Meta's large stock offering tied to AI spending. Any TPU order from OpenAI would draw on the same kind of budget.
For now, the careful reading is narrow: a credible outlet says OpenAI plans to buy Google chips, the companies are not commenting, and the hiring trail suggests the work is real but young. Follow the rest of the chip race on our AI page.
Share this story
Spotted an error? Tell us


