Meta Platforms has become one of Microsoft’s largest artificial intelligence customers, spending hundreds of millions of dollars a year for access to AI models through the Azure cloud platform, Bloomberg News reported on Thursday, August 20. The report, attributed to a person familiar with the arrangement, points to a major commercial link between two companies that also compete for users, engineers and influence in the global AI market.
The source told Bloomberg that Meta consumes trillions of tokens each week through Azure. Tokens are the units used to measure the text and other data processed by AI models, so the reported volume indicates extensive use of rented computing and model services. Bloomberg said the person requested anonymity because the details were internal, while Meta and Microsoft both declined to comment. The companies therefore have not independently confirmed the spending figure, the duration of the arrangement or which models Meta accesses.
The relationship is notable because Meta is simultaneously building its own models, data centers and computing capacity. Renting additional AI services from Microsoft can give the social-media group flexible capacity while its internal infrastructure expands, although neither company explained the specific workloads involved. Microsoft, meanwhile, has broadened Azure’s model catalogue beyond its close partnership with OpenAI, positioning the cloud platform as a marketplace for several model providers and corporate customers.
Microsoft reported in July that annual Azure revenue had exceeded $100 billion for the first time. Its quarterly cloud revenue reached $59.3 billion, up 27 percent from a year earlier, according to company results covered by AP News. Chief executive Satya Nadella said at the time that customer confidence was driving the company’s AI and cloud growth. The newly reported Meta usage offers a concrete example of the large corporate demand behind those aggregate figures, though Microsoft has not disclosed revenue from this customer.
The report also arrives during an exceptional infrastructure spending cycle. Axios reported after the companies’ latest results that Meta expected 2026 capital expenditure of $130 billion to $145 billion, while Microsoft’s quarterly capital expenditure rose 70 percent to $41 billion to support cloud and AI demand. Those investments show that leading technology companies are both constructing their own systems and buying capacity from one another as demand for model training and everyday AI inference rises.
The disclosure highlights how interdependent the AI industry has become even among direct rivals, with model developers, cloud operators and consumer platforms exchanging access to scarce computing resources. Important commercial details remain unknown, and neither company has announced a formal contract. Investors and customers will now watch future earnings statements and regulatory filings for confirmation of the scale, cost and strategic purpose of Meta’s reported Azure use.
Comments