Meta and Microsoft Cut Internal Claude AI Use

Meta Platforms and Microsoft are reducing their internal reliance on Anthropic’s Claude AI models as both technology giants encourage employees to adopt proprietary artificial intelligence tools, according to a report from The Information.

Microsoft had earlier projected that its internal spending on Anthropic technology would reach at least $1 billion this year, according to people familiar with the company’s budget. That estimate has since been reduced by more than one-third as Microsoft looks to control operating expenses.

Company leadership has reportedly encouraged employees to limit their use of Claude and shift toward Microsoft’s own AI products. However, the change applies specifically to internal operations. Customer demand for Anthropic models offered through Microsoft’s enterprise platforms continues to grow steadily.

Meta is making a similar transition. Internal use of Anthropic’s Claude Code coding assistant has fallen to around 30,000 employees from approximately 60,000 earlier this year.

Layoffs during the spring, which affected roughly 10% of Meta’s 78,000 employees, contributed to the decline. However, Meta’s growing portfolio of internally developed AI models and coding tools appears to be the main factor reducing its dependence on Claude.

The company has increasingly promoted its proprietary AI ecosystem, including its Muse Spark family of models. Meta began externally testing Muse Code in August, positioning the product as a direct competitor to Claude Code. The Meta-powered coding assistant has already attracted more than 6,000 internal users.

MetaCode, another AI coding tool developed specifically for Meta employees, has expanded even faster and now has more than 30,000 internal users.

The shifts at Microsoft and Meta highlight a broader push among major technology companies to reduce dependence on third-party AI providers while investing heavily in proprietary models and software. Anthropic remains an important AI provider, particularly for enterprise customers, but increasingly capable in-house alternatives could reshape how major technology firms allocate their AI spending.