Anthropic is committing $30B in Azure compute, powered by Nvidia chips, to scale Claude β and in return receives $5B from Microsoft and $10B from Nvidia. Same cycle, same actors, different roles: provider, client, investor β sometimes all three at once.
The effect is immediate. Anthropic was valued at $183B in September and today sits at $350B. Part of that jump reflects real growth β enterprise adoption, agent demand, competitive positioning against OpenAI. But a significant part is explained by the agreement itself, and that distinction matters.
When your investors are also your providers and your buyers, valuation stops measuring only external demand and starts incorporating how much they're willing to spend on each other. The industrial logic is real:
π Anthropic needs the compute to train and scale frontier models.
π Microsoft needs a leading model to compete in the enterprise AI market.
π Nvidia needs customers with enough capital to keep buying its chips.
Each actor needs what the other offers. But Wall Street knows that discomfort around a potential AI bubble doesn't disappear with announcements at this scale. Sometimes it only adds to it.
A snake that bites its own tail can still grow. The question remains the same: is there anyone outside the circle watching?