In a strategic move to optimize spending and promote internal innovation, both Meta and Microsoft are reportedly asking their employees to dial back their usage of Anthropic’s AI model, Claude. The directive encourages staff to lean more on proprietary AI tools developed in-house, a shift aimed at significant cost reduction.
Cost-Cutting Measures Drive AI Tool Shift
Microsoft, in particular, had anticipated substantial expenses related to its use of Anthropic’s AI, with internal projections earlier this year suggesting an annual spend of at least $1 billion on the technology. This forecast prompted an initial directive for employees to reduce their reliance on Claude and prioritize in-house solutions. Recent budget adjustments have reportedly seen this allocation slashed by over a third.
It’s important to note that this internal adjustment by Microsoft does not reflect a decline in demand for Anthropic’s models from its external enterprise clients. The tech giant continues to offer Claude and other AI models through its robust cloud platform, indicating the shift is primarily focused on internal operational efficiencies rather than a broader strategic withdrawal from the partnership.
Meta Sees Significant Drop in Claude Users
Meta’s internal data paints a similar picture, revealing a sharp decrease in the number of employees using Claude Code. The user base has reportedly shrunk to around 30,000, a significant drop from approximately 60,000 earlier this year. While Meta’s overall employee count has decreased due to recent layoffs, the company’s aggressive promotion of its own AI tools is cited as a primary driver for the reduced adoption of Claude.
Meta has been actively developing and deploying a suite of internal AI tools for its workforce. MetaCode, one such tool, has already garnered over 30,000 internal users. Furthermore, Muse Code, which is currently undergoing external client testing, was launched in August and has attracted more than 6,000 internal users. Muse Code is built upon Meta’s proprietary large language models and is widely seen as a direct competitor to Claude Code.
Continued Investment Despite User Decline
Despite the noticeable reduction in internal users for Claude Code, Meta’s expenditure on the platform remains considerable. In the past 28 days, the company reportedly spent over $105 million on Claude Code. This highlights the substantial costs associated with large-scale AI model usage, even as companies pivot towards internal solutions.
Anthropic’s own IPO filing revealed that two major clients accounted for approximately 25% of its revenue. While not explicitly named, Meta and Microsoft are widely speculated to be these key customers. The filing also indicated that most large clients do not have long-term agreements, leaving room for potential shifts in spending.
The revelations underscore a broader trend in the tech industry where companies are increasingly scrutinizing the costs associated with third-party AI services. As the capabilities of in-house AI development mature, major players like Meta and Microsoft are strategically leveraging their own resources to manage expenses and foster internal AI ecosystems.
