Anthropic's public IPO filing has put its rapid AI growth and steep costs in focus as investors assess a proposed valuation of $1.5 trillion to $2 trillion. If achieved, that would rank the company among the most valuable US tech firms. The potential listing—possibly one of the largest ever—also comes as interest rates rise and markets scrutinize the scale of spending on AI infrastructure, making the company difficult to value.
Enterprise customer growth, but revenue remains concentrated
Founded in 2021 by researchers who left OpenAI, Anthropic sells its Claude models and related services mainly through enterprise contracts and subscriptions. Its filing says more than 11,900 business customers spent at least $100,000 with the company over the past year. More than 1,900 spent at least $1 million, while about 250 spent at least $10 million.
That growth has not removed concerns about reliance on a small group of customers. In the first nine months of 2026, five customers accounted for roughly a quarter of revenue. Investors who are positive on the company say securing enterprise business at this scale so early in its development points to strong demand. Denny Fish, a portfolio manager at Janus Henderson Investors, described the technology as a potentially far-reaching turning point, while stressing that the filing's details matter.
Erica Brescia, managing director at Redpoint Ventures, another Anthropic investor, said the company's business performance has been impressive. She said an IPO could give retail investors a way to participate directly in the development of AI companies, rather than limiting their exposure to chipmakers.
Losses and computing commitments complicate valuation
Anthropic recorded a loss of $80.7 billion in the first half of 2026, up from $12.9 billion in the same period of 2025. The filing says most of the loss came from changes in the fair value of convertible notes and derivative liabilities. Investors will therefore need to distinguish accounting valuation changes from the company's underlying operating performance. At the same time, Anthropic's computing commitments are growing, while competition from lower-cost open-weight models is intensifying.
David Trainer, chief executive of New Constructs, remained cautious after reviewing the filing. He pointed to the company's competitive position, arrangements for insider share sales and commitments and liabilities that far exceed its cash. AI may be a powerful tool for society, he said, but that does not necessarily mean companies in the sector can build equally durable business models. Trainer also said investors need more clarity on insider ownership and plans to sell shares.
Roger McNamee, an early investor in Google and Facebook, voiced stronger reservations about the offering. He said the industry still has to prove much of its commercial value and questioned whether a lengthy filing would be enough to help investors understand the relationship between Anthropic's business prospects and its risks. Enthusiasm for the company has also surfaced on social media, but a positive narrative does not replace scrutiny of the financial details.
Pricing and profitability remain key IPO questions
The public filing is only one step in the IPO process. Anthropic still needs to meet investors on a roadshow, gauge demand, set an offer price and allocate shares. Once those steps are complete, the company is expected to list on Nasdaq under the ticker symbol “ANT.” Its valuation was $965 billion in May 2026, well below the proposed IPO range. The final offer price will shape how investors weigh the company's growth against its risks.
Gil Luria, a managing director at D.A. Davidson, said he still wants to understand Anthropic's long-term path to profitability and its eventual valuation. A pure-play AI listing could attract investors if it is priced appropriately, he said. For now, the filing's disclosures on customer concentration, the sources of losses, computing commitments and competition remain central to assessing whether growth can translate into sustained profits.