Meta Platforms has quietly become one of Microsoft's largest artificial intelligence customers, a development that underscores how demand for emerging AI technology remains heavily concentrated within the tech industry itself.
According to a person familiar with the matter, Meta spends hundreds of millions of dollars annually to access AI models through Microsoft's Azure cloud services. The same source indicated that Meta processes trillions of tokens—the standard unit for measuring AI computing consumption—through this platform on a weekly basis.
A key component of Microsoft's AI strategy involves offering models from various vendors through its marketplace platform known as Foundry. As of July, Foundry had attracted 100,000 customers. While Microsoft's promotional materials frequently showcase clients from traditional sectors such as manufacturing and transportation, insiders reveal that the company's largest AI customers today remain predominantly other technology firms.
Beyond Meta, other major clients include Adobe, the AI company Perplexity, and Sierra—a customer-service AI startup co-founded by Bret Taylor, who serves as chairman of OpenAI. Microsoft's reliance on a small cluster of tech-sector clients is particularly pronounced, with its long-time partner OpenAI contributing roughly 70% of Microsoft's total AI revenue in the most recent fiscal year.
In addition to providing model access, Microsoft also offers its AI assistant Copilot and rents out AI-focused computing capacity. Meta, for its part, is investing heavily in AI to support software development, purchasing model access through multiple platforms based on availability and cost considerations. Sources say Meta developers have used OpenAI's technology via Foundry to help evaluate the output of Meta's own in-house models.
Meta's Chief Technology Officer Andrew Bosworth explained during a July appearance on the "Big Technology" podcast that, alongside developing proprietary AI models, the company also rents leading third-party models as part of its development workflow. Simultaneously, Meta is building its own business to offer clients access to a range of AI models through API services, a move that could eventually position it as a competitor to Foundry and help reduce Meta's spending on external providers like Microsoft.
This pattern is not new for Meta, which has previously relied on Microsoft technology before replacing it with in-house solutions. From the late 2000s, Microsoft's Bing search engine powered web searches for Facebook, until Meta phased out that partnership by the end of 2014. In the years before ChatGPT ignited the current AI boom, Microsoft also supplied computing power to Meta for AI development. Since then, Meta has become one of the world's largest builders of AI data centers to support its model development efforts.
The fact that Microsoft's biggest AI clients remain other tech companies is not encouraging news for some investors. For AI to truly meet market expectations, the technology must achieve widespread adoption across the entire economy, not just among technologically sophisticated firms—especially given ongoing concerns within the tech sector about revenue concentration and the potential drawbacks of circular business dealings between companies.
Comments