Anthropic has reassured investors that its adjusted operating income will remain positive for another quarter, according to people familiar with the matter. The company is also expected to post gross margins of more than 80% before accounting for model-training costs and revenue-sharing arrangements with partners such as Amazon.

The reported results come as Anthropic faces questions over the cost of developing artificial intelligence systems and prepares for a potential initial public offering. Sustaining profitability would represent an important milestone for the company as investors increasingly assess whether leading AI firms can convert rapid revenue growth into durable earnings instead of relying on continued access to large amounts of capital.

Anthropic has reportedly selected Nasdaq for a possible public listing, which could value the company at more than $2 trillion. If the listing proceeds at that level, it could surpass the record set by SpaceX, which went public in June at a valuation of $1.77 trillion, according to a Reuters report.

The company had been expected to publish its prospectus last week. Instead, it has reportedly given a selected group of investors access to the documents and is collecting their feedback before making them public.

Positive earnings would strengthen Anthropic’s case ahead of a potential listing, particularly if the company can continue reporting adjusted operating income while expanding its customer base. However, its operations also involve substantial costs related to computing capacity, model development and commercial partnerships. The combination of high gross margins and continued heavy investment in AI systems could influence how public-market investors assess the company’s long-term valuation.

Revenue growth and forecasts

Anthropic achieved adjusted operating profit in the second quarter on revenue of $11.5 billion, representing a 14-fold increase from the same period a year earlier. Its momentum continued into late July, when its annual revenue run rate reached $65 billion, well above the $9 billion baseline recorded at the end of last year. The company continues to anticipate a profit in the third quarter.

Joey Brookhart, an AI lab analyst at SemiAnalysis, said some investors expect Anthropic’s annual revenue to reach $120 billion by the end of the year and to nearly triple that amount by the end of 2027. Additional reports said the company is forecasting revenue of approximately $190 billion to $200 billion in 2028.

Anthropic said it already has 6,000 customers that each spend at least $100,000 annually. Brookhart cautioned, however, that the company’s current margins could make it difficult for other firms to compete. He said, “If you continue to operate at these margins and growth rates, it will be so hard to compete [with Anthropic] because they have so much [computing resource].”

The company’s progress may become more difficult to assess as calls grow for a slower pace of AI development. A pause in development could save Anthropic billions of dollars in model-training expenses, but it could also allow competitors to narrow the gap.

Chief Executive Officer Dario Amodei has argued that the industry should slow the pace of model upgrades as concerns increase over the power of AI systems. In a statement published on his website, he said, “Not building the technology deprives humanity of benefits or simply places AI in the hands of authoritarian powers, while building it too fast is reckless.” The full statement is available in Amodei’s post.

Amodei said that drastic improvements in AI, if left unmanaged, could exceed society’s ability to understand or control the systems. He argued that development must therefore proceed with extreme caution, if it proceeds at all.

Sam Altman and Elon Musk have expressed similar positions. Altman told Fortune that OpenAI will remain private this year despite filing a confidential IPO document in June, saying that 2026 is not an appropriate time for the company to go public.

Several AI firms have also held private discussions in recent weeks about mutual safety protocols. Sources close to those talks said they were prompted by three factors: recent cybersecurity breaches, increasing concern among researchers about next-generation AI capabilities, and the political reality that President Donald Trump is unlikely to stop the development of AI.

Demand for Anthropic’s most expensive model

Most major AI companies are relying on new models to drive future profits. Yet data from Ramp’s tracking of corporate expenses indicates that spending on Anthropic’s most expensive and powerful model, Fable 5, has stalled at approximately 11% of total sales, despite the model having been available for more than two months.

Anthropic’s lower-priced Opus 5 model also surpassed Fable 5 in business popularity shortly after its launch in late July.

Several large AI laboratories are investing billions of dollars in attempts to build increasingly capable models, based on the assumption that customers will pay premium prices for superior performance. If enterprise buyers instead favor less expensive systems that satisfy their basic requirements, the economic assumptions underlying the industry could change significantly.

Anthropic’s financial position therefore presents a mixed picture ahead of a potential IPO. Strong revenue growth, high gross margins and an expanding enterprise customer base could support the company’s case for a multitrillion-dollar valuation. At the same time, slowing demand for its most expensive model underscores the challenge of persuading customers to pay more for increasingly advanced AI capabilities.

As Anthropic and its competitors continue to spend heavily on computing and model development, investors will be watching whether rapid growth in AI can produce sustainable profits. The company’s ability to balance expansion with rising development costs could become one of the central factors shaping any public-market debut.