Big tech is quietly backing away from Anthropic's Claude — and the timing couldn't be worse.
$META and $MSFT are both actively pushing employees off Claude and onto their own internal AI tools. Microsoft is slashing internal Claude spending by over a third. Meta's Claude usage among its code teams has been cut in half as the company doubles down on its own models.
This is the supply-chain detective story in reverse: when your biggest customers are also your biggest competitors, you're not in a moat business — you're in a temporary licensing deal. Anthropic doesn't control the distribution. It doesn't own the enterprise relationship. And now the landlords are kicking it out of the building.
The IPO rumor mill has been spinning for months, but this is a terrible setup. Prospective investors will ask the obvious question: if your anchor customers are actively reducing dependency, what's the growth story? Anthropic's edge was always supposed to be safety, reasoning quality, and enterprise trust — but if the enterprises with the most AI budget are building in-house, that edge evaporates.
Watch how this plays in the private markets. If Anthropic tries to go public in this environment, the valuation will get marked down hard. The comp isn't OpenAI anymore — it's every other AI infra company that got squeezed between hyperscaler buildout and open-source commoditization.
$META and $MSFT are both actively pushing employees off Claude and onto their own internal AI tools. Microsoft is slashing internal Claude spending by over a third. Meta's Claude usage among its code teams has been cut in half as the company doubles down on its own models.
This is the supply-chain detective story in reverse: when your biggest customers are also your biggest competitors, you're not in a moat business — you're in a temporary licensing deal. Anthropic doesn't control the distribution. It doesn't own the enterprise relationship. And now the landlords are kicking it out of the building.
The IPO rumor mill has been spinning for months, but this is a terrible setup. Prospective investors will ask the obvious question: if your anchor customers are actively reducing dependency, what's the growth story? Anthropic's edge was always supposed to be safety, reasoning quality, and enterprise trust — but if the enterprises with the most AI budget are building in-house, that edge evaporates.
Watch how this plays in the private markets. If Anthropic tries to go public in this environment, the valuation will get marked down hard. The comp isn't OpenAI anymore — it's every other AI infra company that got squeezed between hyperscaler buildout and open-source commoditization.