Meta and Microsoft Curb Internal Use of Claude AI Amid Cost and Workflow Concerns
Meta and Microsoft are significantly reducing their employees' use of Anthropic's Claude AI, prioritizing internal coding tools and AI solutions to manage costs and streamline developer workflows.
Major technology firms Meta and Microsoft are reportedly scaling back their internal reliance on Anthropic's Claude AI, a move driven by a confluence of factors including escalating artificial intelligence costs and a strategic push towards proprietary developer tools. The Information reported on October 5th that both companies are encouraging employees to shift their focus to in-house solutions like Microsoft's GitHub Copilot and Meta's own MetaCode and Muse Code platforms.
This strategic pivot is not an outright rejection of Claude, as both Meta and Microsoft remain substantial customers of Anthropic. Instead, the adjustments primarily target internal spending and the optimization of developer workflows. The increasing integration of AI coding assistants into daily software development tasks has brought the associated costs under intense scrutiny, prompting these large organizations to re-evaluate their AI expenditure and toolchain strategies. The evolving landscape sees AI spending, product roadmaps, and control over developer environments becoming increasingly intertwined.
Microsoft, in particular, has reportedly revised its internal AI spending projections. Previously anticipating annual spending on Anthropic technology to exceed $1 billion, the company has since seen this figure fall by over a third. This reduction is attributed to directives for employees to decrease Claude usage and increase reliance on Microsoft's own AI offerings, such as GitHub Copilot and various OpenAI models. Furthermore, internal monthly AI spending limits within Microsoft's cloud and AI division have been drastically lowered, from $100,000 per employee to approximately $10,000 in many cases, a move that has reportedly caused some frustration among engineers accustomed to broader exploratory budgets.
It is crucial to distinguish between internal usage and customer-facing products. While Microsoft is tightening its employees' access to Claude, the AI models remain available and are reportedly seeing continued customer spending growth through Microsoft's enterprise platforms. This dual approach highlights a complex strategy of managing internal resources while continuing to leverage external AI capabilities for its client base.
Meta has also observed a significant decrease in Claude Code users, dropping from around 60,000 earlier in the year to approximately 30,000. While workforce reductions have played a role, Meta's deliberate promotion of its own AI tools, including MetaCode and Muse Code, is cited as the primary driver. These internal tools have garnered substantial employee adoption, with MetaCode boasting over 30,000 users and Muse Code over 6,000. Meta has even begun testing Muse Code with external customers since August, signaling a growing confidence in its proprietary AI solutions.
Despite the reduced user numbers, Meta's spending on Claude Code remains considerable, with reports indicating over $105 million spent in a single 28-day period. This underscores that even with fewer users, the cost of advanced AI tools can remain high, reinforcing the financial rationale behind the shift.
From a security perspective, the transition between AI coding assistants does not eliminate the inherent risks associated with automated tools accessing sensitive data, commands, and credentials. Past vulnerabilities in Claude Code, such as those allowing unauthorized execution and API key theft due to repository misconfigurations, and similar issues with Microsoft's tools like the RoguePilot vulnerability, emphasize the ongoing need for robust security controls. These include stringent permission management, the use of trusted project settings, and diligent oversight of AI-generated actions.
Anthropic's continued strong market performance, with a reported annualized revenue pace of $65 billion, indicates sustained demand for its AI services. The decisions by Meta and Microsoft to curb internal usage reflect a broader industry trend towards cost optimization and strategic alignment of AI tools with core business objectives, rather than a fundamental loss of faith in Claude's capabilities. The focus remains on balancing innovation with fiscal responsibility and security.