Anthropic expects positive adjusted operating income for a second consecutive quarter, according to people familiar with the matter, as the AI startup prepares for a potential public offering.
The company's gross margins exceed 80 percent before accounting for revenue shared with distribution partners like Amazon and expenses tied to training its Claude models, sources said.
The distinction matters. Anthropic's reported margin strips out two of the largest cost buckets in frontier AI economics: the wholesale revenue splits it owes channel partners and the ongoing GPU and compute infrastructure required to train and fine-tune models. For a company that will compete on marginal unit economics, showing net margins—profit after those deductions—would tell a sharper story.
Achieving adjusted operating profitability for two consecutive quarters represents a turning point for an AI developer that has prioritized research spending over near-term returns. Most frontier AI labs still run negative operating margins when training costs are included. Anthropic's shift suggests either that inference-driven revenue (API calls, subscriptions) is scaling faster than expected, or that the company has achieved material efficiency gains in model inference costs—likely both.
That progression matters for IPO valuation. Investors in legacy software watch for gross margins in the 70–80 percent range as a sign of pricing power and distribution leverage. For Anthropic, the question is whether that core margin can survive the margin compression from partner payouts and whether the company can eventually show positive net operating income without carving out training expenses.
Anthropic has attracted backing from Salesforce Ventures, Google, and others, but it remains capital-intensive relative to mature software businesses. The company's path to sustained profitability depends on whether it can grow API revenue and commercial deployments faster than the cost of maintaining frontier-grade model training.
