Anthropic is seeking a valuation of about $1.3 trillion in its initial public offering, the company behind the Claude chatbot said in a regulatory filing on Tuesday. It plans to sell 825 million shares priced between $70 and $78 each, which would let the San Francisco firm raise more than $61 billion if the offering is priced at the midpoint of that range.

The company is expected to make its Wall Street debut this month, listing on the Nasdaq Global Select Market under the ticker “ANT”. At that size it would rank among the biggest IPOs ever, though it now looks set to fall short of the record $86 billion raised by SpaceX in June. Going public would also put Anthropic ahead of rival OpenAI, setting up a pair of blockbuster debuts by competing AI developers.

Pricing a listing is a delicate exercise. Set the price too high and the shares can slide on the first day. Set it too low and the company and its banks leave money on the table.

A regulatory filing last month showed Anthropic’s revenue growing rapidly, passing $16 billion in the first half of the year, 14 times its figure for the same period a year earlier. Like many tech companies heading to the market, it has yet to turn a profit. Anthropic, OpenAI and SpaceX’s AI division have all been losing more than they make because of the heavy cost of building AI, fuelling talk of a bubble.

That talk sharpened after a brutal sell-off on Wall Street late last week, led by stocks tied to the construction of AI data centres. Shares in AI companies have swung for months on worries that further data centre building could slow as borrowing costs climb.

Formed in 2021 by former OpenAI leaders, Anthropic is locked in a contest for dominance with that company, alongside Google, newer startups in the United States and China and SpaceX. The listing could add billions of dollars to the fortunes of chief executive Dario Amodei and his sister and co-founder, Daniela Amodei.

In September, Dario Amodei cautioned that the industry should slow its development to let safety measures catch up. Without that, he warned, AI could let loose a swarm of agents capable of taking over the entire internet within six to 12 months. In a post on his website he set out a plan for tighter checks, saying Anthropic is already doing part of this itself but that the rest would require coordination across the industry and with governments around the world, including authoritarian ones.