Anthropic's IPO sets new benchmark
Valuation underpinned by expanding base of paying enterprise customers
By LI JIAYING | China Daily | Updated: 2026-11-20 09:38
US artificial intelligence heavyweight Anthropic's record-setting initial public offering is providing investors worldwide with a new public-market benchmark for valuing AI companies, while also sharpening the focus on how Chinese large language model firms can turn growing demand into sustainable earnings, industry experts said.
The Claude developer set its Nasdaq offer price at $83 on Wednesday, above its initial range of $70 to $78, raising $81.3 billion and reaching a fully diluted valuation of about $1.5 trillion. Including a separate $26 billion private placement, its total fundraising is set to reach about $107.3 billion, surpassing the $85.7 billion raised by previous record holder SpaceX's debut in June.
"Anthropic's valuation reflects a combination of its overall model capabilities, customer quality, revenue growth and long-term capital expectations," said Guo Tao, an angel investor and senior AI expert.
Providing a new public-market valuation benchmark for AI enterprises worldwide, its valuation is underpinned by a rapidly expanding base of paying enterprise customers, sustained revenue growth and the effectiveness of its models in production environments, Guo said.
"Chinese companies can draw on the core yardsticks of revenue growth, value per customer and returns per unit of computing power, focusing on real, paying applications rather than simply competing on parameter counts," he said.
These benchmarks are particularly relevant as listed Chinese model developers increasingly seek to monetize their technologies through application programming interface (API) services and enterprise applications.
Hong Kong-listed Zhipu AI reported revenue of 954 million yuan ($143.1 million) in the first half, up 399.7 percent year-on-year, with open-platform and API services accounting for 86.5 percent of the total.
MiniMax recorded a top line of $116.6 million, up 283.1 percent, with open-platform and other AI-based enterprise services contributing 63.4 percent.
Unlike US frontier model developers that place greater emphasis on premium model capabilities and long-term enterprise subscriptions, Chinese peers typically combine API services with customized solutions and private deployments for sectors such as manufacturing and government services, Guo said.
Their value is increasingly assessed in terms of application efficiency, end-to-end cost controls and industrial collaboration, rather than technological capabilities alone, he added.
However, strong revenue growth has yet to translate into comparable profitability. Zhipu and Mini-Max reported interim gross profit margins of 26.4 percent and 17.9 percent, respectively, significantly below Anthropic's reported 85.4 percent over the same period.
The challenge is also reflected in how much of their research spending Chinese AI developers can cover through gross profit alone, said Beijing-based tech analyst Poe Zhao, in an earlier interview. In the first half, Anthropic's gross profit covered its research and development spending about 1.4 times, compared with roughly one-eighth for Zhipu and one-fourteenth for MiniMax.
"Low prices can win usage, but at today's margins they cannot fund research," Zhao said, highlighting the need for Chinese developers to demonstrate stronger pricing power.
Improving margins will require companies to move beyond price competition in standardized AI services, according to a Nov 11 report by SPDB International. While routine tasks such as general question-answering and text editing face growing price pressure, advanced models can still command a premium in complex coding and long-running AI agent tasks, where greater accuracy and reliability can reduce human intervention and rework, it said.
The report identified two key drivers of better profitability — optimizing inference and expanding usage volumes to lower unit costs and lift gross margins; and allowing revenue growth to outpace R&D spending, gradually reducing R&D costs as a share of revenue.
lijiaying@chinadaily.com.cn




















