Anthropic, the San Francisco developer of the Claude artificial intelligence models, ended its first week as a public company on Friday with its shares still well above their $83 offer price. The result consolidated a debut that market reports said valued the company at about $1.76 trillion. Its stock held most of the gains it made in its first session.

The company's shares began trading on Thursday. They opened at $111.50 and closed at $97.17, about 17 per cent above the price set by the company and its underwriters. Friday's trading was the first test of whether early buyers would take quick profits, and the stock stayed comfortably above its listing level into the weekend.

The performance handed sizeable paper gains to Amazon and Alphabet, Google's parent company. Both invested billions of dollars in Anthropic over the past three years. Amazon has committed about $18bn to the company and supplies much of its computing power through Amazon Web Services. Google has also taken a significant minority stake, and both companies hold the investments on their balance sheets at values that the listing has now made public.

Bankers and fund managers have treated the listing as a gauge of public-market appetite for loss-making frontier AI developers. Rivals including OpenAI have been weighing their own routes to market. A steady first week strengthened the case that investors will pay for AI model makers on revenue growth alone, despite the heavy spending on chips and data centres those companies need.

The listing came as the broader market slid, even though Nasdaq rose at the open, after Nvidia reported record quarterly revenue earlier in the week. Anthropic's lock-up arrangements, which restrict early investors and staff from selling shares for a set period after the offering, mean most insider stock cannot yet reach the market. Analysts at the banks that underwrote the deal are barred from publishing research until the standard quiet period ends.