Meta and Microsoft Slash Anthropic Claude Usage, Pivot to In-House AI Tools

Meta and Microsoft Scale Back Internal Use of Anthropic's Claude

Meta and Microsoft are reducing reliance on Anthropic Claude as both companies steer staff from Anthropic Claude toward in-house AI tools1

1

3

3

. Microsoft projected at least $1 billion in internal spending on Anthropic's technology earlier this year but has since slashed that estimate by more than one-third2

2

4

4

. Meta cut Claude Code users from approximately 60,000 earlier this year to 30,000, largely by pushing employees toward proprietary AI solutions including Muse Code and MetaCode2

2

4

4

. The AI industry shift reflects cost considerations and strategic control as companies prioritize building over licensing at hyperscale operations.

Cost Considerations Drive Strategic Pivot to In-House AI

AI spending has become difficult to forecast, with just 11% of almost 400 surveyed businesses able to accurately project their AI costs2

2

. The top 1% of U.S. businesses spent a median of $7,400 per employee on AI in July, while the median company spent just $11.95 per employee2

2

. Meta was on track to spend billions of dollars this year on internal use of AI tools before imposing token limits and spending controls1

1

. Microsoft's internal Claude spending included Claude Code, Claude models in Copilot, and Claude Mythos2

2

. Building in-house AI tools requires bigger investment upfront, but data stays internal and tools can be shaped around actual employee workflows3

3

. Companies operating at Meta and Microsoft's scale find that internal tools running on their own infrastructure become easier to justify when improving fast enough to compete with outside alternatives.

MetaCode and Copilot Gain Traction as Proprietary Alternatives

MetaCode, Meta's internal coding assistant, has already surpassed 30,000 users inside the company1

1

4

4

. Meta began external testing of Muse Code in August, which had reached more than 6,000 internal users3

3

4

4

. Microsoft has directed employees to prioritize GitHub Copilot, which has reached 50 million users with growing paid Copilot seats3

3

. A Microsoft spokesperson confirmed the company has been steering staff to use GitHub Copilot coding tool, although engineers can still choose other models2

2

. Companies using their own AI ecosystems generate usage data that helps those tools improve, creating feedback loops unavailable when licensing external models3

3

. Meta's spring layoffs, which cut approximately 10% of the workforce, account for part of the Claude Code decline, but the primary driver remains the push toward in-house AI tools3

3

4

4

.

Competitive Dynamics and National Security Concerns Impact Anthropic

Anthropic captured 43.5% of U.S. businesses paying for its subscriptions or tokens as of July, leading overall business adoption2

2

. However, Anthropic was designated a national security supply chain risk by the U.S. Department of Defense, altering consumer and enterprise perception regardless of the designation's potentially political founding1

1

. Anthropic warned in an IPO prospectus that "the company may experience material revenue losses or business disruptions attributable to these events"1

1

. Losing billions in revenue from two of its biggest customers could seriously impact a company slated to go public for a reported valuation of over $2 trillion1

1

. Enterprise demand for Claude through Microsoft's platform has held steady, and customer spending on Anthropic models via Microsoft's enterprise platforms continues to see steady growth3

3

4

4

.

What This Means for AI Tooling and Enterprise Adoption

Microsoft can integrate its own models more directly into GitHub Copilot, Microsoft 365, and Azure, while Meta can train internal tools on workflows specific to how its engineers build and ship products3

3

. Companies consolidating real advantage are those owning cloud compute, developer tools, and enterprise software layers3

3

. Hyperscalers with engineering capacity to build their own AI tools will eventually do so, while enterprises that cannot build for themselves represent more durable demand for providers like Anthropic3

3

. Meta has been in discussions about a compute deal with Anthropic, where Anthropic would pay for access to Meta's data center infrastructure, showing commercial relationships and competitive dynamics can exist simultaneously3

3

. Watch for continued shifts as companies weigh the economics of building versus licensing, and monitor how Anthropic's enterprise customer base evolves as hyperscalers develop competing solutions.