SI Agent News
DevelopingClaudeBusiness

Microsoft and Meta rein in staff use of Claude as Anthropic's IPO nears

Anthropic's two biggest business customers are trimming internal Claude use just before its stock listing, and a commentary asks whether price is the reason.

The short version

Why it matters

The story shows that a top-rated AI tool can still lose ground when the bill gets large. If you buy AI services, expect more scrutiny of usage, spending caps and cheaper alternatives. Treat the piece as opinion built on reported figures, not as proof that Anthropic is in trouble.

Two exits and two budget cuts

As tmtpost.com reported in a commentary, two prominent members of the Claude Code team, Thariq Shihipar and Anthony Morris, announced they were leaving within an hour of each other. Neither said where they were going. The timing stood out because Anthropic's planned stock listing was only about three weeks away.

The departures are the visible part. The piece leans more heavily on an earlier report from The Information: Microsoft and Meta, Anthropic's two largest business customers, are limiting how freely their own employees use Claude.

What the two giants changed

Microsoft had expected its internal spending on Anthropic technology to reach at least one billion dollars a year. That budget was reportedly cut by more than a third, and staff were pushed toward GitHub Copilot and in-house models. The monthly per-person cap on AI model spending in one division was reportedly lowered by about nine-tenths in most cases.

At Meta, the number of employees using Claude Code roughly halved. Part of that follows spring layoffs, but the commentary says the bigger driver is Meta's push for its own tools. Even so, Meta still spent more than a hundred million dollars on Claude Code in the latest 28 days, so the bill barely shrank.

Not a cancellation

The commentary stresses that neither company has dropped Claude. Microsoft's purchases of Claude for features it sells to outside customers are reportedly still growing, and both firms still resell Claude through their cloud platforms. What changed is that they no longer let their own staff use it without limit.

The reason, the author argues, is that Claude Code spread by word of mouth inside Microsoft. A token (a small chunk of text a model reads or writes, and the unit it bills by) costs add up quickly. Per engineer, the cost was reportedly several hundred to two thousand dollars a month, so AI tools now get budgets and audits like cloud servers.

Why it lands before an IPO

Anthropic's draft listing materials reportedly show revenue growing more than ten-fold in six months, alongside a heavy net loss. They also say that nearly a quarter of 2025 revenue came from two customers, and that many large customers lack long-term contracts. The media did not name the two, but the commentary suggests Microsoft and Meta fit.

Some investors have reportedly lowered their fair-value estimate from about two trillion to about one and a half trillion dollars. This is the commentary's reading of events, not a company statement.

What to watch

Try it

Look at your own AI bill this week. List which tasks use the most, set a monthly cap where the tool allows it, and test whether a cheaper model handles your simple jobs.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by tmtpost.com · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news