Anthropic has announced that its AI model Claude is now profitable, claiming a 20% operating margin on its subscription and API services. However, the company's financial disclosures omit significant costs, including the massive capital expenditures for training next-generation models. Independent analysts estimate that when factoring in R&D and infrastructure, Anthropic is still deeply unprofitable. The announcement appears designed to attract further investment ahead of a potential IPO.
Profitability is a magic trick. Anthropic waves a hand, shows you a rabbit. But the hat is full of holes. They count revenue from Claude. They ignore the billions spent on compute. They pretend training costs don't exist. This is not profit. This is a story for investors.
Tech companies love this game. Burn cash for years. Declare victory on a technicality. The media eats it up. But the numbers don't lie. Anthropic is a money furnace. The only profit is in the narrative. And narratives collapse.