SAN FRANCISCO — OpenAI plans to buy Google's artificial intelligence chips, The Information reported on Tuesday, 27 October 2026, in a move that would deepen an increasingly tangled relationship between two of the world's most powerful AI rivals — competitors in products and search, partners in computing.
According to Reuters, which reported The Information's story on 28 October, the planned purchase would see OpenAI buy chips from Alphabet's Google, citing a person familiar with the matter. The report did not disclose financial terms or other details, and neither company confirmed the deal: Google did not immediately respond to a Reuters request for comment outside regular business hours, while OpenAI declined to comment. But if completed, the purchase would mark a striking evolution — the company behind ChatGPT, one of the largest buyers of Nvidia's graphics processors, buying silicon from one of its fiercest rivals.
Illustration: Two different processor chips facing each other on a circuit board — an artist's impression, not a photograph of any company's hardware.
What Happened
The Information's report, published on Tuesday and carried by Reuters on Wednesday, says OpenAI plans to purchase Google's AI chips as technology companies ramp up spending on data centres and semiconductors to meet surging demand for AI computing. The chips in question are Google's Tensor Processing Units, or TPUs — processors designed by Google for AI workloads, which have gained traction as an alternative to Nvidia's market-leading graphics processing units (GPUs) as companies seek cheaper options and chips better suited to running AI models once they are built.
What makes the report significant is not only the buyer and seller, but the direction of travel. OpenAI and Google compete head-on: OpenAI's ChatGPT against Google's Gemini, OpenAI's search ambitions against Google's search dominance. Yet the economics of building and deploying large AI models are pushing them together. Last week, OpenAI said it would make its frontier AI models available through Google Cloud, and Reuters reported in June 2025 that OpenAI had already begun renting Google's AI chips. A purchase would take that relationship one step further, from renting capacity to owning hardware.
The timing is also telling. Alphabet was due to report its quarterly results later on Wednesday, 28 October, with investors watching for signs that its heavy investment in AI is continuing to drive growth in its cloud business. Deals with customers including rival Anthropic have already fuelled demand for Google's TPUs, and Alphabet's chief financial officer, Anat Ashkenazi, has previously said the company expects to begin recognising a small percentage of revenue from TPU sales agreements by the end of this year, with the vast majority expected in 2027. An OpenAI purchase would add the industry's most prominent name to that customer list.
It is important to be precise about what is known and what is not. The purchase is reported, not announced. The Information cited a single person familiar with the matter, according to Reuters, and neither OpenAI nor Google has confirmed it. Financial terms, quantities and timelines are all undisclosed. Readers should therefore treat the deal as a credible report from a well-sourced technology publication, not as a completed transaction — while recognising that the surrounding facts, from the cloud partnership to the chip rentals, are already established.
Background
To understand why this matters, it helps to understand the chip bottleneck at the heart of the AI boom. Training a frontier AI model requires thousands of specialised processors running for weeks or months; running those models for millions of users — known as inference — requires vast fleets of chips operating around the clock. For years, Nvidia has dominated both markets, and OpenAI has been one of its largest customers, using Nvidia GPUs to train its models and run inference computing.
Google's TPUs offer a different path. Designed in-house and deployed across Google's own data centres for years, TPUs are optimised for the kind of large-scale AI work that Google itself performs, and they have increasingly been pitched as a cheaper, more available alternative to Nvidia hardware. Google began selling TPUs directly to some customers this year, a shift from its earlier practice of offering them mainly through its cloud service. That commercial opening is what makes an OpenAI purchase possible at all: Google is now in the business of selling chips to outsiders, including rivals.
Illustration: Rows of server racks in a vast data centre corridor — an artist's impression, not a photograph of any company's facility.
The OpenAI-Google relationship has been deepening for months, even as the two companies compete. In April 2026, OpenAI's exclusive distribution arrangement with Microsoft expired, freeing it to sell its models through other cloud providers. It has since expanded distribution to Amazon Web Services and, most recently, to Google Cloud. Under the Google Cloud deal announced on 21 October 2026, OpenAI's latest models, including GPT-6 Astra, and its Codex coding agent are being made available through Google Cloud's Model Garden, allowing customers to buy them through the Google Cloud Marketplace and use existing cloud spending commitments. For Google Cloud, adding OpenAI's models strengthens a strategy of presenting itself as a neutral platform offering a wide choice of models rather than relying only on its own.
The wider industry is moving the same way. On Tuesday, 27 October, Oracle said it was expanding its cloud AI chip deals with Nvidia to what it called "multi-gigawatt scale" with Nvidia's newest Vera Rubin servers, while also diversifying to offer Google's TPU chips for the first time, alongside chips from Cerebras Systems and SambaNova Systems. Oracle's cloud chief, Mahesh Thiagarajan, said different AI workloads have different compute needs, and that choice gives customers flexibility over performance, throughput and economics. The message from Oracle, OpenAI and Google alike is that the AI chip market is becoming a multi-supplier market, not a single-supplier one.
The sums involved are staggering. BusinessWorld, reporting on The Information's story, noted that Amazon, Microsoft and Google parent 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. Against that background, securing diverse, cost-effective chip supply is not a side issue for AI companies — it is one of their central strategic problems.
Why It Matters
The reported deal matters first because it signals a loosening of Nvidia's grip. Nvidia remains dominant, and OpenAI remains one of its largest buyers, but every major AI company now has an incentive to diversify its chip supply — for price, for availability and for bargaining power. If OpenAI, the most prominent AI company in the world, buys Google's chips at scale, it validates TPUs as a serious alternative and encourages other buyers to follow. That, in turn, could reshape the economics of the whole AI hardware market, putting competitive pressure on prices and accelerating innovation.
It matters second because it illustrates how competition and cooperation now coexist in AI. The old model of technology rivalry — two companies fighting over the same market with no overlap — does not describe OpenAI and Google. They compete in chatbots, search and enterprise AI, yet they sell each other cloud services, distribute each other's models and, if the report is accurate, will soon trade hardware. This "coopetition" is driven by the sheer cost of AI infrastructure: no company, however large, can build everything alone, and partnerships between rivals are becoming a structural feature of the industry rather than an exception.
It matters third for Google itself. Selling TPUs to OpenAI would be both a revenue opportunity and a strategic statement. Google has spent years and billions of dollars developing its chip programme, largely for internal use. Turning that programme into a merchant business — selling chips to external customers, including direct competitors — is a major shift, and landing OpenAI as a customer would be its most visible proof point. It would also strengthen Google Cloud's pitch as the neutral home for AI workloads of all kinds, regardless of which company's models a customer uses.
Finally, it matters for the debate about AI spending. Investors have grown anxious about whether the hundreds of billions being poured into data centres and chips will ever earn adequate returns. Deals like this one cut both ways in that debate: they show demand for AI computing is real and growing, but they also show how expensive and complex it is to meet that demand. The question facing the sector, as market commentators have put it, is whether spending on data centres and chips can generate sustainable returns commensurate with its scale. An OpenAI-Google chip deal does not answer that question, but it shows the industry's leading companies are still betting heavily that it can.
What Happens Next
The most immediate next step is confirmation — or otherwise. If OpenAI and Google confirm the purchase, expect details on scale, timing and which TPU generation is involved, and expect analysts to parse Alphabet's earnings commentary for hints about TPU revenue recognition. If the companies stay silent, the report will remain what it is now: a well-sourced but unconfirmed claim, to be weighed alongside the established facts of the OpenAI-Google partnership.
Illustration: Two stylised robotic hands shaking over a cloud symbol — an artist's impression of a partnership between rival technology companies.
Alphabet's quarterly results, due later on Wednesday, are the next scheduled event. Investors will look for growth in the cloud business, for commentary on TPU demand and for any signal about how much of the company's AI infrastructure spending is translating into customer revenue. The results will also set the tone for the rest of the week's big technology earnings, in a market that has grown sensitive to any sign that AI spending is outrunning returns.
More broadly, watch for further diversification deals. Oracle's move to offer Google, Cerebras and SambaNova chips alongside Nvidia's is part of a pattern, and other cloud providers and AI labs are likely to announce similar multi-supplier strategies. The inference market, in particular — the business of running AI models for users, rather than training them — is more competitive and diverse than the training market, and it is there that alternative chips have their best opening.
For OpenAI, the strategic logic is clear: more chip suppliers mean more bargaining power, more supply security and potentially lower costs, all of which matter as it scales its models to hundreds of millions of users. For Google, the logic is equally clear: every external TPU customer strengthens its chip business and its cloud platform. Whether this particular deal closes or not, the forces behind it — cost, scarcity and the need for scale — are not going away. The AI chip market's transformation from a one-supplier world to a multi-supplier one is under way, and this reported purchase, if confirmed, would be one of its clearest milestones.
Sponsored: explore today's featured sponsored offers
FAQ
Has OpenAI actually bought Google's AI chips?
Not confirmed. The Information reported on 27 October 2026, citing a person familiar with the matter, that OpenAI plans to buy Google's AI chips. Reuters carried the report on 28 October, noting that Google did not immediately respond to a request for comment and OpenAI declined to comment. The report disclosed no financial terms or details.
What are Google's AI chips?
Google's Tensor Processing Units (TPUs) are processors designed by Google for AI workloads. They have gained traction as an alternative to Nvidia's market-leading graphics processing units, as companies seek cheaper options and chips suited to running AI models. Google began selling TPUs directly to some customers in 2026.
Why would OpenAI buy chips from a rival?
OpenAI is one of the largest buyers of Nvidia's GPUs, but diversifying chip supply can improve price, availability and bargaining power. The two companies already cooperate despite competing in AI products and search: OpenAI rents Google's chips, and its models are now available through Google Cloud.
What is the OpenAI-Google Cloud deal?
Announced on 21 October 2026, the deal makes OpenAI's frontier models, including GPT-6 Astra, and its Codex coding agent available through Google Cloud's Model Garden, letting customers buy them via the Google Cloud Marketplace. It followed the expiry of OpenAI's exclusive distribution arrangement with Microsoft in April 2026.
What happens next?
Watch for confirmation from OpenAI or Google, for commentary on TPU demand in Alphabet's quarterly results due later on 28 October, and for further multi-supplier chip deals across the industry, following Oracle's announcement that it will offer Google's TPUs alongside Nvidia hardware.
Sources
- Reuters, "OpenAI plans to buy Google's AI chips, The Information reports," 28 October 2026.
- Reuters, "OpenAI to sell its top models on Google Cloud in latest enterprise push," 21 October 2026.
- Reuters, "Oracle expands chip supply deal with Nvidia, adds Google and Cerebras offerings," 27 October 2026.
- BusinessWorld Online, "OpenAI plans to buy Google's AI chips, The Information reports," 28 October 2026.
Comments
Post a Comment