OpenAI has signed a major cloud computing agreement with Google, even though the two companies compete directly in the fast-growing AI industry. The deal was finalized in May 2025 and will help OpenAI meet its rising need for computing power, three sources familiar with the matter told Reuters.
The move comes as OpenAI continues to expand its services, including ChatGPT, which has become a popular AI tool used by millions of people worldwide. To support the demand, OpenAI needs more computer capacity, known in the industry as “compute.” This deal adds Google Cloud to the list of providers helping meet that demand.
The partnership is unusual because OpenAI and Google are rivals in the AI space. OpenAI’s ChatGPT is widely seen as a major threat to Google’s dominance in online search. Even so, both companies see the need to cooperate due to the massive amount of computing power needed to train and run AI models. This is especially true as ChatGPT use grows across businesses, schools, and individual users.
Until early 2025, OpenAI relied mainly on Microsoft Azure for its computing needs. Microsoft is a major investor in OpenAI. But OpenAI has recently moved to expand its cloud partnerships. Earlier this year, it signed big contracts with CoreWeave and joined forces with Oracle and SoftBank on the large Stargate infrastructure project. These steps are part of OpenAI’s plan to avoid depending too much on any single partner.
In February, reports showed that OpenAI is also working on building its own custom chips. This could help the company reduce its need for chips from other suppliers and improve the speed and cost of running its models. The deal with Google is the latest example of how OpenAI is spreading its infrastructure across more partners.
Alphabet, Google’s parent company, saw its stock rise by over 2 percent after the news. Meanwhile, Microsoft’s shares dropped slightly. Analysts from major banks said the deal was surprising, but also a big win for Google’s cloud unit. At the same time, they warned that ChatGPT could still pose a serious risk to Google’s core search business.
The cloud deal also shows how Google is working to expand its cloud services to AI-focused startups. Google Cloud earned $43 billion in revenue in 2024, which made up about 12 percent of Alphabet’s total earnings. The company is now selling access to its special AI chips, called TPUs, to outside clients. These chips were once used only for Google’s own AI research.
By selling computing power to OpenAI, Google may be helping a rival. But it also shows that cloud services are now a key part of the tech industry’s growth. Other AI firms, like Anthropic and Safe Superintelligence, have also signed with Google Cloud. These startups were started by former OpenAI team members and are now building competing AI systems.
Google’s DeepMind unit is also in the race to develop better AI models. This means Google must now divide its computing resources between its own teams and external customers like OpenAI. The company’s CFO recently said that Google already has more demand than it can supply. Still, adding OpenAI to its list of clients shows how serious Google is about becoming a leader in AI infrastructure.
OpenAI, meanwhile, has reached a revenue run rate of $10 billion as of June. This means the company is likely to meet its full-year financial goals, driven by growing demand for AI tools across industries.
The deal also comes as Microsoft and OpenAI are discussing new terms for their multi-billion-dollar partnership. These talks include changes to Microsoft’s future stake in OpenAI, which could affect how closely the two companies work together in the coming years.
While the partnership between OpenAI and Google may seem unusual, it shows how the AI industry is evolving. Companies that compete in some areas may still choose to work together when infrastructure and capacity are at stake. As demand for AI continues to rise, more of these types of partnerships are likely to happen.
