Microsoft trims internal Claude spend by a third as Meta pulls back too

Microsoft is cutting its internal spending on Anthropic's Claude by a third, and Meta is moving to reduce how much its employees rely on the model at all. The Information reported the shift, with Cointelegraph flagging it as breaking news. Two of the largest names in tech quietly stepping back from a rival's AI is the kind of move the industry notices fast.

Claude is Anthropic's flagship model family, and it has been a common plug-in for companies that want frontier AI without building it in-house. Now the emphasis is on reliance — how much day-to-day work runs through someone else's model — and spending is the first lever Microsoft is pulling.

Neither company framed the pullback as a full exit, based on the report. The thread tying the two together is internal use: employees reaching for Claude to get work done, and leadership deciding that dependence should shrink.

Both firms already have their own horses in the AI race, which makes a retreat from an outside model less about capability and more about control. Anthropic, meanwhile, has built much of its business on enterprises and developers renting its models rather than owning them.

What fills the gap — and how quickly — is the open question. Watch whether the pullback stays contained to internal tooling or spreads, and whether other big employers follow Microsoft and Meta's lead.

For Anthropic, the stakes are simple: its biggest selling point is that the world's largest companies trust Claude enough to build on it. A reported downshift at two of them is the kind of headline that gets read closely.