OpenAI is turning to one of its fiercest AI competitors for more computing power. According to Reuters, the company has struck a deal to use Google Cloud infrastructure, a move that broadens OpenAI’s AI compute supply beyond Microsoft Azure.
The agreement was finalized in May after months of negotiations. It comes as OpenAI’s need for data center capacity has grown alongside its model training, product operations, and reported annual revenue of $10 billion as of June.
A major shift in OpenAI’s cloud strategy
For years, Microsoft was the central cloud partner behind OpenAI’s expansion. Microsoft’s relationship with OpenAI dates back to 2019, with significant expansions of investment from the computer giant in 2021 and 2023.
That arrangement gave OpenAI a major infrastructure backer as ChatGPT and its related AI systems became widely used. But the source article says Microsoft Azure had been OpenAI’s exclusive cloud provider only until January. The Google Cloud deal marks a clear change from relying on one cloud provider alone.
The reason is straightforward: OpenAI needs more computing capacity. Under the new deal, Google Cloud will provide additional infrastructure to help OpenAI train and run its AI models.
That distinction matters. Training AI models requires large amounts of compute, while running models for users also demands reliable infrastructure. OpenAI’s decision to add Google Cloud suggests that capacity, speed of access, and diversification have become strategic issues, not just back-end technical concerns.
Why rivals are working together
The deal is striking because Google and OpenAI compete directly in artificial intelligence. Google’s DeepMind AI unit is a direct competitor in developing AI models and consumer applications. OpenAI’s ChatGPT also poses what the source describes as arguably the most serious threat to Google’s search business in years.
Even so, the partnership shows that the AI market is not only a contest between isolated companies. Competitors can also become customers, suppliers, and infrastructure partners when both sides see an advantage.
For OpenAI, the advantage is more AI compute. For Google, OpenAI becomes a major Google Cloud customer at a time when cloud providers are competing to serve AI companies with heavy infrastructure needs.
The source also notes that Google’s CEO recently suggested the AI race may not be winner-take-all. The OpenAI-Google Cloud deal fits that idea: a company can compete in AI models while still selling computing resources to another leading AI lab.
Microsoft remains central, but not exclusive
The Google Cloud agreement does not erase Microsoft’s role. Microsoft and OpenAI are also negotiating to revise their multibillion-dollar investment terms, including Microsoft’s future equity stake in OpenAI.
What has changed is the exclusivity. Reuters reported that Google and OpenAI had discussed an arrangement for months but were previously blocked from signing because of OpenAI’s exclusive arrangement with Microsoft.
That detail helps explain the timing. Once OpenAI was no longer limited to Azure alone, it could pursue other data center and cloud options. The Google Cloud deal is one result of that broader infrastructure strategy.
The source article also points to earlier reporting from The Information that OpenAI had begun seeking data center deals elsewhere. The reason cited was the need for more AI data center servers faster than Microsoft could supply them.
A broader race for AI infrastructure
The Google Cloud deal is not the only sign that OpenAI is trying to expand its computing foundation. The company’s recent infrastructure efforts also include the nominally $500 billion Stargate project with SoftBank and Oracle, as well as billion-dollar deals with CoreWeave.
OpenAI is also developing its first in-house chip. According to the source, that work could reduce dependence on external hardware providers like Nvidia.
Taken together, these moves point to a single pressure: AI companies need vast computing resources to train and operate their systems. Cloud contracts, data center deals, specialized chips, and in-house hardware plans all become part of the same strategic puzzle.
For readers following the AI industry, the most important point is that model quality is only one part of the race. The companies building and operating AI systems also need access to physical infrastructure, chips, and cloud capacity at scale.
What Google gains from the deal
For Google, adding OpenAI as a customer supports its broader cloud strategy. Reuters notes that Google has used its in-house tensor processing units, or TPUs, to grow its cloud business.
These specialized chips are described in the source as somewhat similar to the GPUs produced by Nvidia to accelerate AI tasks. They have helped Google attract customers, including Apple and Anthropic.
Google Cloud generated $43 billion in sales comprising 12 percent of Alphabet’s 2024 revenue. The business has positioned itself as a neutral provider of computing resources, aiming to challenge Amazon and Microsoft as the preferred cloud provider for AI startups with heavy AI compute needs.
There is a tradeoff. Selling computing power to OpenAI reduces Google’s own chip supply while potentially strengthening a rival. But the deal also signals that Google Cloud wants to be an infrastructure provider even for companies competing with Google’s own AI products.
That is what makes the arrangement important. It is not just another cloud contract. It is a sign that the AI arms race is being shaped by partnerships that cross competitive lines, because the demand for compute is large enough to make rivals useful to one another.