OpenAI Reportedly Plans to Buy Google’s AI Chips as AI Infrastructure Spending Nears $600 Billion

OpenAI is planning to buy Google’s AI chips, according to a report published on October 28, 2026 — a move that would deepen one of the most unusual partnerships in technology, between two companies that compete fiercely in AI products and search while increasingly doing business with each other. The report, from The Information and carried by Reuters, says OpenAI would purchase Google’s Tensor Processing Units (TPUs), the custom AI accelerators Google has developed as an alternative to Nvidia’s market-leading graphics processing units (GPUs).

Neither company has confirmed the reported purchase. OpenAI declined to comment on the report, and Google did not immediately respond to a request for comment. That caution matters: this is a reported plan attributed to a person familiar with the matter, not an announced deal with disclosed terms. The report did not disclose financial terms or other details. What is confirmed, however, is the context around it — a context that makes the report credible and strategically significant whether or not a purchase is ultimately announced.

Just last week, OpenAI said it would make its frontier AI models available through Google Cloud. Reuters reported in June 2025 that OpenAI had begun renting Google’s AI chips. And Google’s TPUs have been gaining traction with customers, including rival Anthropic, as companies look for cheaper alternatives to Nvidia hardware. Taken together, those facts describe a relationship that has already moved from pure rivalry to practical cooperation — and a chip purchase would be the next step in that direction.

What Is Reported, and What Is Confirmed

Precision matters with a story like this, because the difference between a reported plan and a signed deal is the difference between a signal and a fact. The table below separates the two.

OpenAI’s decision to decline comment, rather than deny the report, is also worth noting — though it proves nothing on its own. Companies routinely decline to comment on unannounced plans. The test of this report will be whether a purchase, a supply agreement, or a related cloud arrangement is announced in the coming weeks or months.

Why OpenAI Would Buy Google’s Chips

The core reason is supply, cost, and leverage. OpenAI is already among the largest buyers of Nvidia’s GPUs, which it uses to train and deploy its AI models. Nvidia’s hardware dominates AI training, but that dominance comes with constraints: high prices, long lead times at peak demand, and dependence on a single supplier for the most critical input to OpenAI’s business. Adding a second major supplier — Google’s TPUs — reduces that dependence and gives OpenAI negotiating leverage with both vendors.

Cost is the second factor. Google’s TPUs have gained traction precisely because companies are seeking cheaper options than Nvidia’s market-leading chips. For inference — running trained models to serve user queries, as opposed to training them — TPUs can be cost-competitive, and inference is where the bulk of ongoing compute cost sits as AI products scale to hundreds of millions of users. If OpenAI can serve ChatGPT queries more cheaply on TPUs, the savings compound across every query.

The third factor is capacity. Demand for AI computing is surging faster than any single supplier can comfortably meet. Amazon, Microsoft, and Google parent Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure, according to the reporting around this story. In that environment, securing chips from every credible source is not a luxury — it is a necessity. A company that waits for its preferred supplier risks waiting in a queue behind competitors.

GPUs vs TPUs vs Custom Accelerators: The Hardware Comparison

The AI chip market is no longer a one-horse race. Three broad approaches now compete, and OpenAI’s reported move touches all three.

Google began selling TPUs directly to some customers this year, a shift from its earlier approach of offering them mainly through Google Cloud. That change is what makes a direct purchase by OpenAI plausible: TPUs are no longer only a cloud service, but a product that outside companies can buy. Alphabet expects to begin recognizing a small percentage of revenue from TPU sales agreements by the end of this year, with the vast majority expected in 2027, according to comments from Alphabet CFO Anat Ashkenazi cited in the reporting. That timeline suggests TPU sales are at an inflection point — small now, but structured to grow quickly.

The Anthropic Parallel

OpenAI would not be the first AI lab to lean on Google’s hardware. Deals with customers including rival Anthropic have fueled demand for Google’s TPUs, and Anthropic’s own trajectory illustrates the scale of the infrastructure race: AI shares extended their rebound on October 27 after Anthropic filed for a US listing, and a group of banks and Blackstone committed $63 billion to finance computing equipment for the company. That financing vehicle will buy Broadcom accelerators and networking systems for lease to Anthropic, with deliveries expected to begin in 2027.

The Anthropic financing also illustrates the other side of the story: the AI rally has not settled questions about how the data-centre build-out will be funded, or whether it will earn returns. Investors are weighing the demand outlook against the debt and lease risks behind AI infrastructure expansion. A $63 billion financing structure for one company’s equipment is, by any measure, an extraordinary bet on future AI revenue — and it is a bet financed with debt, not equity alone.

The $600 Billion Infrastructure Question

The reported OpenAI-Google chip plan sits inside a much larger spending wave. Amazon, Microsoft, and Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure, as investors increasingly question when such spending will translate into meaningful returns. That question — when does the spending pay off? — is now the central debate in technology investing.

There are two ways to read the spending. The optimistic reading is that AI infrastructure is like railroad or telecom infrastructure: expensive to build, but once built, it supports decades of profitable services. On that reading, today’s spending is a down payment on a platform shift, and the companies that build the most capacity will capture the most value. The sceptical reading is that much of the spending is defensive — each company spending because its rivals are spending — and that returns will be thinner and later than the market expects, especially where capacity is financed with debt rather than cash flow.

Alphabet’s quarterly results, due later on October 28, will be watched for signs of which reading is closer to the truth in Google’s case: whether heavy AI investment is continuing to drive growth in its cloud business, and how TPU-related revenue is trending. Those results will not settle the debate, but they will add one more data point to it.

What This Means for the AI Industry

For OpenAI

A TPU purchase would diversify OpenAI’s compute supply, potentially lower its inference costs, and strengthen its bargaining position with Nvidia. It would also deepen its dependence on Google — a competitor in search and AI products — which is a strategic risk OpenAI must weigh against the supply benefit. The fact that OpenAI already rents Google chips and distributes models through Google Cloud suggests it has already decided that the benefit outweighs the risk.

For Google

Selling TPUs to OpenAI would be a landmark validation of Google’s chip program. Google has spent years and billions developing TPUs; selling them to the company behind ChatGPT would signal that TPUs are competitive outside Google’s own workloads. It would also create an awkward but profitable dynamic: Google supplying the hardware that powers a rival’s flagship product, while competing with that product through Gemini.

For Nvidia

OpenAI is one of Nvidia’s most important customers. Losing even a portion of OpenAI’s future orders to Google would not dethrone Nvidia, which remains dominant in training, but it would confirm that the era of Nvidia as the only credible supplier is ending. Nvidia’s response — on price, on supply, and on its own product roadmap — will shape the next phase of the chip race.

For Everyone Else

For smaller AI companies and developers, the OpenAI-Google story is good news in one respect: a more competitive chip market should eventually mean cheaper, more available compute. But in the short term, the largest companies are absorbing most of the available capacity, financed by sums — $600 billion in combined infrastructure spending, $63 billion in a single financing vehicle — that smaller firms cannot match. The gap between the AI haves and have-nots is widening, even as the underlying hardware becomes more diverse.

What to Watch Next

Several milestones will clarify whether this report becomes a deal. First, any formal announcement from OpenAI or Google about a chip supply agreement. Second, Alphabet’s quarterly results later on October 28, particularly commentary on cloud growth and TPU revenue. Third, Anthropic’s US listing process and the deployment of its $63 billion equipment financing from 2027. And fourth, Nvidia’s response — whether in pricing, supply commitments, or new product announcements — to the prospect of its largest customers diversifying.

None of these will arrive on a predictable timetable, and the absence of an announcement in the next few weeks would not disprove the report: large chip deals are often negotiated over months and announced only when hardware is ready to ship. What is certain is the direction of travel. The AI industry’s compute supply chain is diversifying, Google’s TPUs are a serious part of that diversification, and even direct competitors are now willing to buy from each other when the alternative is waiting in line.

Frequently Asked Questions

Is OpenAI buying Google’s AI chips?

According to a report from The Information on October 28, 2026, OpenAI plans to buy Google’s AI chips. The report has not been confirmed by either company: OpenAI declined to comment and Google did not immediately respond. No financial terms were disclosed.

What are Google’s AI chips called?

Tensor Processing Units, or TPUs. They are custom AI accelerators developed by Google as an alternative to Nvidia’s GPUs, and Google began selling them directly to some customers this year.

Why would OpenAI buy chips from a competitor?

To diversify its compute supply, potentially lower inference costs, and secure capacity in a market where demand is surging. OpenAI already rents Google chips and distributes its models through Google Cloud, so a purchase would extend an existing commercial relationship.

How much are tech giants spending on AI infrastructure?

Amazon, Microsoft, and Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure, according to reporting around this story. Separately, a $63 billion financing vehicle will buy Broadcom equipment for lease to Anthropic.

Does this mean Nvidia is losing its lead?

Not yet. Nvidia remains dominant, particularly in AI training, and OpenAI remains one of its largest customers. But the reported plan confirms that major customers are actively building second sources of supply, which is a meaningful change in the chip market’s structure.

Reporting note: the chip-purchase plan is attributed to The Information via Reuters/BusinessWorld reporting dated October 28, 2026, and is presented here as a report, not a confirmed deal. Confirmed context — OpenAI’s Google Cloud distribution, the June 2025 chip-rental report, and Alphabet’s TPU revenue commentary — is identified as such above.