Anthropic reports $17B in Q3 revenue, commitments of $895B in revised S-1

The thread treated the filing less as a victory lap for AI demand than as a stress test for whether frontier-model revenue can ever justify frontier-model spending. Commenters agree that Anthropic has built an astonishing business quickly: a $17B quarter and roughly $68B annualized revenue would have been implausible for an AI company not long ago. Even the slower quarter still implies 47% sequential growth, and optimists argue that a company scaling from a tiny base should not be valued like a mature software vendor. The mood turns sharply cautious once costs and durability enter the picture. Readers focused on compute costs rising faster than revenue, margins flipping negative, enormous future infrastructure commitments, and a possible need to finance cash shortfalls expensively. They argued that an IPO valuation near $2T requires not merely continued growth but hundreds of billions in eventual revenue and unusually strong margins. Several see model access as a commodity: customers can route workloads elsewhere when prices or quality move, while open-weight models keep pressing on hosted API pricing. Others countered that broker-platform token shares are a poor window into enterprise demand, where direct contracts and still-unsettled internal AI deployments may matter far more. Accounting-related losses were debated too, but few treated them as enough to erase the underlying cash-flow and dilution risks.