Anthropic Profit Claim Comes With Big AI Cost Questions

Anthropic is reportedly telling investors a better financial story, but the details show why AI profitability is still a slippery thing to measure.

The Financial Times reported that Anthropic has achieved profitability for a second straight quarter on an adjusted operating income basis, a claim that could help reassure investors as the Claude maker prepares for a possible public listing. That is a big shift for a company operating in one of the most expensive technology markets in history.

The caveat is the word adjusted. MarketWatch noted that the profitability metric excludes stock-based compensation and does not fully reflect some of the most painful costs in frontier AI, including training expenses and revenue-sharing arrangements with partners.

This does not mean Anthropic’s progress is fake. It means investors have to read the numbers with unusual care. A company can show strong gross margins on services while still needing enormous capital to train future models, rent cloud capacity, buy inference compute and support enterprise growth.

That distinction matters because Anthropic is no longer just a research lab with a good chatbot. It is becoming a major enterprise AI company with large cloud partnerships, heavy compute needs and growing expectations around a future IPO. If the company can show durable revenue growth and improving margins, investors will take it seriously. If the profits depend too heavily on adjustments, the market will ask harder questions.

The timing also creates an interesting contrast. Anthropic is talking about financial strength at the same moment its CEO is warning that frontier AI may need to slow down for safety. TechBooky recently wrote about how Nvidia’s reported Anthropic IPO talks raise circular AI finance questions. The new profitability report adds another layer to that debate.

For the AI industry, the bigger lesson is that profitability will not be judged like ordinary software. Investors will want to know how much revenue comes from real customers, how much goes back to cloud and chip partners, how much future training will cost and whether safety delays could affect product timelines.

Anthropic may be moving toward a stronger financial position. But in frontier AI, profit is not just a number on a slide. It is a claim about whether the whole business model can survive the cost of staying at the frontier.