First Pour: is Anthropic over-leveraged?
Anthropic is going public and I read the S-1 so you don’t have to.
Here's the Tea;LDR:
Revenue hit $16.3 billion in the first half of 2026, up 1,300% year over year. Extraordinary.
The reported net loss was $80.7 billion, but most of that came from accounting adjustments, not cash leaving the business.
The real Tea IMO: Anthropic had $266 billion in minimum purchase commitments and $321 billion in future lease payments, primarily for compute and infrastructure.
That’s nearly $600 billion committed to the bet that AI demand keeps soaring.
Here’s where it gets interesting: Anthropic relies on Amazon, Google and Microsoft for compute. Those same companies pay to use Claude, help distribute Claude, and offer competing AI models of their own.
So… Anthropic’s suppliers are also their customers…are also their distribution partners…are also their competitors.
That's a lot of money owed and revenue tied to your frenemies.
Meanwhile, cheaper open-weight models keep getting better. What happens if Anthropic’s biggest customers decide “good enough” is actually, good enough?
My Tea: There’s no doubt that Anthropic has built an extraordinary, historic business. But the competitive pressure are real; and the strategic questions are too.
AI demand isn’t a moat. The technology isn't one either. So then, what exactly is? Something to brew on. ☕️