Editorial illustration for Meta, Microsoft slash Claude spending as user base halves
Meta and Microsoft Cut Claude Spending as Users Drop
Meta, Microsoft slash Claude spending as user base halves
Microsoft had been on pace to spend more than $1 billion a year on Anthropic's Claude, according to The Information. That number is now shrinking fast. Cloud division executives Scott Guthrie and Jay Parikh told staff to switch to GitHub Copilot and OpenAI's models instead, and the monthly per-employee budget for Claude dropped from $100,000 to roughly $10,000, a cut of more than a third overall.
Meta is making a similar move. The number of employees using Claude Code has fallen from about 60,000 to 30,000, a slide tied partly to layoffs and partly to Meta's push to get people onto its own Muse Code and MetaCode tools. The scale of the prior spending was enormous: in one 28-day stretch, Meta put more than $105 million into Claude Code alone. The company had already started tightening AI costs back in June.
Both companies have been among Anthropic's largest enterprise customers, which makes the pullback notable beyond simple budget trimming. Anthropic isn't just a vendor to these firms anymore. It's a company building products that increasingly look like direct competition to what Microsoft and Meta are trying to sell.
Why this matters
Anthropic sold Claude to Microsoft and Meta as infrastructure. Both companies are now treating it as a line item they'd rather not pay, and the numbers show how fast that calculus flipped. A cloud division budget dropping from $100,000 to $10,000 per employee isn't trimming, it's a near-total retreat.
Meta halving its Claude Code user base tells a similar story, even accounting for layoffs. For developers and founders building on top of frontier models, the lesson is that even the deepest-pocketed customers will walk the moment they have a credible in-house alternative, GitHub Copilot, OpenAI's models, Muse Code, MetaCode. Anthropic's revenue base was never going to stay reliant on companies that see it as a rival as much as a vendor.
Researchers tracking the model wars should watch whether Claude's enterprise pullback shows up in Anthropic's next funding pitch or pricing moves. The bigger pattern: platform companies don't stay customers of AI labs they're racing against. They become competitors, then they stop paying rent.
Related Reading
Browse all Anthropic news →
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
Related Topic Claude gains shared context in Excel, PowerPoint; Microsoft adds Copilot Cowork
AI Tools & Apps
Related Topic Anthropic: Hackers Stole Claude Tokens Via Compromised Session Key
LLMs & Generative AI
From Archives Instinct's AI Agent Enters Group Chats, Bypassing Account Requirement
From Archives Sam Altman says AI risks are worth the benefits
From Archives Anthropic's Opus 5.5 Boosts Speed, Lowers Price in AI Podcast Rundown
AI Tools & Apps
From Archives Claude Code's New Mods System Lets Developers Rewrite the AI Tool Internally
LLMs & Generative AI
Common Questions Answered
Why did Microsoft reduce its monthly per-employee budget for Claude from $100,000 to $10,000?
Microsoft's cloud division executives Scott Guthrie and Jay Parikh instructed staff to switch to GitHub Copilot and OpenAI's models instead of Claude. This represents a cut of more than a third overall in Claude spending, with Microsoft previously on pace to spend over $1 billion annually on Anthropic's Claude service.
How did Meta's Claude Code user base change and what does this indicate?
Meta's Claude Code user base was halved as the company made a similar retreat from Anthropic's services as Microsoft. This significant reduction demonstrates that both major tech companies are treating Claude as a line item they would rather not pay for, even accounting for layoffs.
What are the motivations behind Meta and Microsoft's pullback from Anthropic beyond cost savings?
Beyond cost savings, Meta has little incentive to continue funding Anthropic when it wants to sell competing products and reportedly sought to restrict Anthropic's access to its training data. This shift reflects how Anthropic has transformed from being treated as infrastructure and partner to being viewed as a competitor by both companies.
What is the significance of this spending reduction for developers building on frontier AI models?
The near-total retreat by major companies from Claude spending demonstrates how quickly the calculus around AI model partnerships can flip, suggesting that developers and founders should be cautious about building heavily on any single frontier model. The lesson indicates that even strong partnerships with tech giants can rapidly deprioritize alternative AI services when strategic interests diverge.
