Meta and Microsoft are trying to reduce employee reliance on Claude: Meta's internal user base has been halved, and Microsoft's budget has been cut by a third.

Anthropic's revenue outlook is under pressure as Meta and Microsoft work to reduce employee use of Claude.

On October 5, The Information reported that Meta and Microsoft are significantly reducing their internal use of Anthropic's Claude AI, instead encouraging employees to adopt their own self-developed tools.

This shift directly impacts Anthropic's core revenue structure, which the company disclosed in its recent IPO prospectus, stating that its two major customers together contribute approximately 25% of its revenue.

Microsoft had previously projected annual internal spending on Claude to be at least $1 billion, but that budget has been cut by more than a third. Meanwhile, the number of employees within Meta using Claude Code has plummeted from approximately 60,000 earlier this year to around 30,000, a drop of nearly 50%.

These adjustments are significant for Anthropic. While both tech giants emphasize that demand for Claude from external clients through enterprise platforms continues to grow, the decline in internal usage indicates that large tech companies are accelerating the in-house integration of AI capabilities.

Microsoft: Cutting internal Claude budget and shifting to in-house model development

The core logic behind Microsoft's move is to control operating costs.

According to reports citing sources familiar with the matter, Microsoft began reducing its reliance on Claude Code as early as May of this year, with the aim of reducing employee token consumption and promoting the use of internal models.

Microsoft had previously projected that its internal spending on Claude AI would reach $1 billion, but that budget has been cut by more than a third after employees were asked to reduce their use of Claude to save costs and spend more time on Microsoft’s own AI tools.

It is worth noting that this adjustment is limited to internal operations, and the Anthropic model service provided to external customers through the Microsoft Enterprise Platform continues to grow steadily.

Meta: Self-developed tools accelerate development, substitution effect emerges.

The decline in Claude usage within Meta is due to both personnel factors and strategic intentions.

This spring, Meta laid off about 10% of its approximately 78,000 employees, which to some extent reduced Claude's user base. However, the more fundamental driving force, according to reports, lies in the rapid maturation of Meta's own AI ecosystem.

Meta has launched two internal programming assistant tools: MetaCode, which is exclusively for internal use and currently has more than 30,000 users; and Muse Code, which began testing with external customers in August and currently has more than 6,000 internal users.

Muse Code is considered a direct competitor to Claude Code. It is built on Meta's self-developed model and is part of its Muse Spark series of model products.

Despite a significant drop in Claude Code's user base, Meta still spent over $105 million on Claude Code in the last 28 days.

Anthropic: Customer concentration risk raises concerns

Anthropic disclosed in its recent IPO prospectus that two clients collectively contributed approximately 25% of its revenue, but did not name the specific clients. The market widely believes that Meta and Microsoft are highly likely to be these two clients or one of them.

The simultaneous reduction in internal usage by two core clients, coupled with the rapid rise of self-developed alternative tools, has created a structural risk of excessive customer concentration for Anthropic during its critical IPO window.

While external deployment demand from enterprises is still growing, whether it can offset the revenue pressure caused by shrinking internal usage remains a key uncertainty for investors.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.