Anthropic posts 50% revenue jump to $16.8bn ahead of landmark IPO

Anthropic told investors that third-quarter revenue rose 50 percent to $16.8 billion and that the company was profitable for a second straight quarter, a disclosure aimed at shoring up support for a blockbuster public listing that will test the durability of the artificial intelligence trade. The Claude maker, which gathered investors in San Francisco on Wednesday for a full day of presentations, is widely expected to launch the most highly valued initial public offering on record, with a reported valuation near $965 billion. The figures, which the company presented ahead of a planned November listing, give Wall Street its clearest view yet of the economics behind one of the most capital-intensive businesses in modern finance.

Revenue in the three months to September rose from $11.5 billion in the prior quarter, though the pace of growth has moderated from earlier in the year, when revenue more than doubled between the first and second quarters, from roughly $4.8 billion. Anthropic said it was cash flow positive and generated $1.70 in revenue for every dollar spent on computing power in the third quarter, with operating margins of 11 percent. The efficiency metrics are central to the company’s pitch: chips and data centres, the infrastructure required to train and run AI models, represent its largest expense, and investors have watched closely to see whether revenue can keep pace with the enormous cost of building out that capacity.

The numbers matter beyond Anthropic itself. Along with rival OpenAI, the company’s revenue trajectory is a crucial gauge of whether the AI boom can justify the hundreds of billions of dollars flowing into chips, data centres and cloud capacity. That question has become more urgent as the macroeconomic backdrop deteriorates. US chip stocks fell sharply on Wednesday on rising inflation and concerns about rampant borrowing to finance the AI buildout, a reminder that the sector’s fortunes remain tied to interest rates and credit conditions. Anthropic has said it plans to spend more than $500 billion on infrastructure and computing power over the next decade, a commitment that becomes harder to sustain if financing costs rise.

The company’s path to the public market is complicated by more than just the numbers. In an early version of its S-1 filing circulated to a small group of investors last month, almost a third of the document was devoted to risk factors, including the possibility that AI models could exhibit unpredictable and malicious behaviours. Anthropic has warned investors that its technology could pose existential risks to humanity, and executives devoted a large portion of Wednesday’s presentation to the company’s safety philosophy. Recent cases of hacking and other misbehaviour by AI models have made it harder for investors to value the company, which must also answer conventional questions about whether it can sustain its growth against competition from OpenAI, Meta and others.

The listing will therefore function as a referendum on two separate propositions. The first is whether Anthropic can keep growing fast enough to justify a valuation that would rank it among the largest companies in the world. The second is whether investors can price a technology whose own maker acknowledges it could pose existential risks. In an early version of its S-1 filing circulated to a small group of investors last month, nearly a third of the document was devoted to risk factors, including the possibility that AI models could exhibit unpredictable and malicious behaviour. The company has warned that its technology could pose existential risks to humanity, and it devoted a substantial portion of Wednesday’s presentation to its safety philosophy.

The listing will therefore be a test of two things at once: whether the market still has appetite for AI companies at extreme valuations, and whether investors can reconcile those valuations with the technology’s known and stated risks. Anthropic’s results suggest the underlying business is improving, but the company faces mounting competition from OpenAI, Meta and others, and its plans to spend more than $500 billion on infrastructure over the next decade leave little margin for error. For a market that has priced AI as a generational opportunity, the November IPO will be the clearest signal yet of whether that conviction survives contact with the balance sheet.