Anthropic could be valued at more than $2 trillion when it lists on the New York Stock Exchange this autumn, a level that would make the Claude developer’s flotation the largest in history, according to the Financial Times.
The artificial intelligence company filed to list in June and the float is expected in September or October. Senior executives have not yet set a target price, the FT reported.
A valuation above $2 trillion would carry Anthropic past SpaceX, which became a public company in June at a value of $1.75 trillion. Investors believe Anthropic’s lead in model capability, together with the speed at which businesses have adopted its tools, justifies the valuation, the FT reported.
The company’s annual recurring revenue, its preferred metric, which extrapolates a full-year figure from a single month’s performance, is expected to reach $100 billion to $120 billion by the end of the year. That compares with $48 billion in May and $9 billion at the start of the year.
Co-founded in 2021 by its chief executive, Dario Amodei, Anthropic secured a valuation of $965 billion in May, making it more valuable than the ChatGPT maker OpenAI. Its models have overtaken OpenAI’s among business users this year.
The run-up to the float has not been smooth. Anthropic this year became the first American company to be designated a supply chain risk by the US government, a ruling that restricts federal agencies and defence contractors from using its models. The company is challenging the designation.
Competition intensifies
Competitive pressure has also increased. Moonshot, a Chinese AI start-up, released its Kimi K3 model last month, which performed on a par with Anthropic’s models on many tasks at a fraction of the cost.
OpenAI has cut the price of its models, leading Anthropic to drop planned price increases for its own.
Azeem Azhar, founder of the research group Exponential View, said some of Anthropic’s recent actions had felt “reactive rather than strategic”.
“Its decision not to raise prices on models like Sonnet felt reactive given the market pressures. It’s facing all this competition from cheaper open-source models, as well as OpenAI’s own price cuts,” he said.
‘Cracks in the AI thesis’
Data from Ramp, a payments company that tracks corporate spending on AI, suggests Anthropic’s market share among US businesses has continued to grow, but that corporate adoption of Fable 5, its most expensive model, has been slow.
According to Ramp, Fable 5 made up only 6 per cent of the tokens businesses purchased from Anthropic over the past month. Despite being Anthropic’s most expensive model by far, it accounted for only 11.4 per cent of spending on the company’s models.
Ara Kharazian, lead economist at Ramp, said the data pointed to “cracks in the AI thesis”.
“With Fable 5, we’ve found a new upper bound for how much businesses are willing to spend on AI,” he said. “More performance is not worth the price tag.”
Azhar said Ramp’s data suggested that many businesses were struggling to identify clear use cases for the most powerful models.
“This technology has evolved faster than the capacity of customers to make sense of it,” he said. “Fable is expensive, but it’s also very powerful, which makes it hard to use. Ordinary companies cannot guarantee that if they’re spending $10 on a million tokens, then they’re getting $11 back. The business case is harder to make.”
