Meta Becomes One of Microsoft’s Largest AI Customers
Bloomberg’s Brody Ford reports that Meta has become one of Microsoft’s largest AI customers, spending hundreds of millions of dollars annually on Azure and consuming trillions of tokens a week, apparently largely for coding assistance. Ford argues that the relationship underscores how Microsoft’s AI revenue remains concentrated among a small group of large technology buyers, despite its broader customer base. Meta’s interest in building an API business could eventually turn a major Azure customer into a more direct supplier of AI services.
Microsoft’s AI revenue is still concentrated among a small group of technology buyers
Brody Ford describes Meta’s spending on Microsoft Azure as evidence that AI demand remains concentrated among a relatively small number of large technology and software companies. Meta has quietly become one of Microsoft’s largest AI customers, paying hundreds of millions of dollars annually for access to Azure and consuming trillions of tokens weekly through the platform, according to Bloomberg’s reporting.
That concentration matters because the promised return on AI investment depends on use spreading beyond the companies building and selling the technology. “For AI to really have the ROI that it’s hyped up to be,” Ford says, “it needs to be used by companies across the economy and disseminate broadly.” Instead, he sees much of the usage and revenue coming from a pool of roughly 10 to 20 large software, digital-native, AI-oriented companies.
Microsoft has emphasized the breadth of its AI customer base, citing 100,000 customers of its Foundry offering and identifying companies such as Levi’s as flagship users. Ford’s point is not that a large customer count is insignificant. It is that customer breadth and commercial concentration can coexist: a platform may serve many organizations while a relatively small group of technology companies accounts for especially consequential usage and revenue.
Meta’s reported spend makes that distinction more visible. Alongside TikTok, Adobe, and OpenAI, Ford places Meta among the large technology buyers that currently matter most to Microsoft’s AI business. The commercial question is therefore not simply whether enterprises are adopting AI tools, but whether spending has diffused enough beyond those buyers to support the scale of investment surrounding the sector.
Their AI business for now remains a tech whale business.
Meta’s role also revives the concern about circular business dealings that Ed Ludlow raises: substantial AI spending is moving among companies that are themselves central participants in the AI market. Ford presents Meta’s Azure use as another data point in that pattern, rather than as evidence that demand has already broadened across the economy.
A hyperscale operator is still buying Azure for a token-intensive workload
Meta operates its own data centers at hyperscale, making the practical question consequential: what is it paying Microsoft for?
Ludlow asks whether the spending is for capacity, software suites, or another service. Ford says that, “as far as we can tell,” coding assistance is the bulk of it. Companies that do a great deal of coding, he adds, can consume large volumes of tokens in that process.
That qualification matters. The reporting does not establish a complete breakdown of Meta’s Azure spend or rule out other uses of Microsoft’s services. But Ford’s account identifies coding assistance as the apparent main workload, and explains why such a use could generate unusually large token consumption even for a company that operates extensive infrastructure of its own.
The relationship is thus not neatly captured by the usual distinction between a cloud provider and a company without its own data centers. Meta has substantial computing operations, yet is also a large buyer of a specific AI service through Azure. Its spending illustrates how demand for AI capabilities can sit alongside, rather than be displaced by, an in-house infrastructure footprint.
Meta’s API ambitions could change the terms of the relationship
Ford connects Meta’s Azure spending to separate reporting that Meta has explored spinning up its own API business. The strategic logic becomes easier to see, he says, when hundreds of millions of dollars are going each year to companies such as Microsoft. An API business could allow Meta to bring some work in-house, reduce vendor spending, and monetize the capability in the process.
That does not establish that Meta is already competing directly with Microsoft in this area. It describes a possible shift: a major customer of an AI platform may also explore becoming a supplier of related capabilities. For Microsoft, Meta’s present demand is valuable; for Meta, that demand may help clarify the economic case for developing and selling more of its own services.
The result is a relationship that combines current dependence with potential future repositioning. Ford’s broader point is that AI is creating new combinations of partnership and rivalry among companies that already have large technology businesses and extensive infrastructure.
It’s very interesting to see how the AI era has forged new alliances and competition between companies.
Neither Meta nor Microsoft commented in response to Bloomberg’s report, Ford says. Microsoft continues to highlight Foundry’s 100,000-customer base, while the reporting identifies Meta as a customer whose scale reinforces the central tension: broad adoption may be growing, but the revenue and usage that matter most still appear concentrated among a small set of technology buyers.