Meta and Microsoft are significantly reducing their internal use of Anthropic’s Claude AI in favor of their own proprietary tools. Together, these two major clients accounted for approximately 25% of Anthropic’s revenue. Microsoft, which previously projected annual internal spending of at least $1 billion on Claude, has now cut its expenditure by more than one-third; Meta has reduced its number of Claude users from around 60,000 to approximately 30,000—a decline of nearly 50%. Microsoft began reducing its reliance on Claude in May of this year, while Meta has launched internal tools such as MetaCode and Muse Code. This shift exposes Anthropic to structural risks stemming from excessive customer concentration just as it enters a critical IPO window.Article author and source: Wall Street Journal
Meta and Microsoft are working to reduce employee usage of Claude, putting pressure on Anthropic’s revenue prospects.
On October 5, according to The Information, Meta and Microsoft are significantly reducing internal use of Anthropic’s Claude AI and instead promoting their own proprietary tools among employees.
This shift directly impacts Anthropic’s core revenue structure, as the company disclosed in its recent IPO prospectus that its two largest customers together accounted for approximately 25% of its revenue.
Microsoft previously anticipated internal annual spending on Claude of at least $1 billion, but this budget has been cut by more than a third. Meanwhile, the number of Meta employees using Claude Code has dropped sharply from approximately 60,000 earlier this year to about 30,000, a decline of nearly 50%.
The above adjustments are significant for Anthropic. Although both tech giants emphasize that demand from external customers using Claude through enterprise platforms continues to grow, the decline in internal usage indicates that major tech companies are accelerating the trend of bringing AI capabilities in-house.
Microsoft: Cutting internal Claude budget in favor of self-developed models
The core logic behind Microsoft's move is to control operational costs.
According to reports citing informed sources, Microsoft began reducing its reliance on Claude Code as early as May this year, aiming to reduce employee token consumption and promote the use of internal models.
Microsoft previously anticipated spending up to $1 billion on internal use of Claude AI, but this budget has been reduced by more than one-third after requiring employees to cut back on Claude usage to save costs and allocate more time to Microsoft’s own AI tools.
It is worth noting that this adjustment is limited to internal operations; the Anthropic model services provided to external customers via the Microsoft Enterprise Platform continue to experience steady growth.
Meta: Self-developed tools accelerate progress, demonstrating substitution effects
Internal usage of Claude at Meta has declined, due to both personnel factors and strategic intent.
This spring, Meta laid off about 10% of its approximately 78,000 employees, which partially reduced the user base for Claude. However, reports indicate that a more fundamental driver is the rapid maturation of Meta’s own AI ecosystem.
Meta has successively launched two internal programming assistant tools: MetaCode, exclusively for internal use, with over 30,000 users; and Muse Code, which began external customer testing in August and currently has over 6,000 internal users.
Muse Code is considered a direct competitor to Claude Code and is built on Meta's proprietary model, part of the Muse Spark series of model products.
Despite a significant decline in Claude Code users, Meta spent over $105 million on Claude Code in the past 28 days.
Anthropic: Concentration of clients raises concerns
In its recent IPO prospectus, Anthropic disclosed that two customers collectively accounted for approximately 25% of its revenue, without naming the specific clients. The market generally believes that Meta and Microsoft are likely these two customers or at least one of them.
Two major core clients simultaneously reduced their internal usage, and the rapid rise of self-developed alternative tools has exposed Anthropic to a structural risk of excessive customer concentration during its critical IPO window.
Although demand for external deployment on the enterprise side continues to grow, whether hedging can offset revenue pressure from declining internal usage remains a key uncertainty for investors.