AI photo: VCG

Chinese AI company DeepSeek has made global headlines for the widespread adoption of its large language model, and that popularity may now be translating into rapid revenue growth. Some Chinese analysts estimate that the star firm's annualized revenue could reach $1 billion this year - nearly double the figure projected just a month ago.

Amid the global AI industry's race toward commercialization, DeepSeek's revenue-generating capability - as a leading representative of Chinese AI companies' open-source approach - has drawn close attention both in China and abroad. Many analysts have described the annualized revenue figure as a "validation moment" not only for DeepSeek, but also for the broader Chinese AI industry's ability to monetize its technology.

Why has DeepSeek's annualized revenue grown so rapidly? What does this rapid growth mean for the commercialization of China's AI industry? And what does this reveal about the long-term potential of Chinese AI companies' open-source approach, compared with the closed-source path pursued by many US AI giants?

To answer these questions, the Global Times conducted extensive interviews with industry analysts, delved deeply into publicly available information on DeepSeek's finances, and analyzed industry reports on the market applications of both Chinese and US AI firms. Collectively, the findings highlighted the rising monetization capability and vast potential of DeepSeek and other Chinese companies, driven by their cost-effectiveness and growing popularity both at home and abroad.

Rapid jump

Notably, the $1 billion annualized revenue figure was also reportedly shared by DeepSeek CEO Liang Wenfeng at a recent meeting with investors. The company now aims to raise 50 billion yuan ($7.45 billion) through a second funding round and is preparing for a listing on the Shanghai Stock Exchange, according to multiple media reports.

DeepSeek did not comment on the media reports when reached by the Global Times.

That number itself may not seem significant, but when compared with the figure from just a month ago, it is highly telling - DeepSeek's annualized revenue could double. What drove such a rapid jump in the span of just one month?

The first factor may have been the company's decision to increase prices by 2.3 to 4.5 times. In mid-August, DeepSeek sharply raised application programming interface (API) charges for its latest flagship models while introducing peak and off-peak pricing. API usage is the predominant source of revenue for AI firms.

However, price hikes alone would not naturally lead to such a rapid jump in annualized revenue, which shows how much a company would make in a year if it continued performing at its current pace; growing demand among users even after the price hike is the key, according to Tian Feng, president of the Fast Think Institute and former dean of the Intelligence Industry Research Institute at Chinese AI software giant SenseTime.

Tian told the Global Times that the jump in DeepSeek's annualized revenue shows that global developers are still willing to pay for DeepSeek's products after the price hike. An annualized revenue figure of $1 billion is not a big sum by the standards of the global large model industry, but it shows one thing: Chinese models have gone through a full process in the global market, from low-price customer acquisition to normal pricing, and customers have stayed, he said. 

DeepSeek's V4.1 Flash, the latest model released by the Chinese AI firm, maintains stable popularity among global users. From September 21 to 27, DeepSeek's V4.1 Flash held the top spot for a second week in a row in terms of weekly usage, with 19.6 trillion tokens, up 24 percent from the previous week, according to the latest data from artificial intelligence model aggregation platform OpenRouter.

A token is the basic unit of text processed by an AI model, and it has become a measure of AI usage and computing consumption.

"It's a validation. Customers' willingness to keep paying after API price hikes is an overall endorsement of Chinese models' capabilities, stability and cost-effectiveness. Looking ahead, this will lead overseas enterprises to treat Chinese models as serious options and evaluate them carefully when choosing LLMs," Tian said.

In addition to the price hike and sustained market demand, gross margin on API access is also key to the rapid surge in DeepSeek's annualized revenue. In this area, DeepSeek holds a considerable advantage compared with other major models. According to media reports, DeepSeek's gross margin on API access reached 82.9 percent in the first seven months of the year, thanks to its efficient infrastructure.

Per a widely circulated transcript of Liang's meeting with investors in July, the DeepSeek CEO said the company aims to reap reasonable profit and expounded that he would be satisfied if costs on server expenditure could be recuperated within 10 months. 

Broader commercialization

"DeepSeek's initial performance has shown that global users recognize the efficiency and value for money offered by the Chinese firm," Xiang Ligang, director-general of the Zhongguancun Modern Information Consumer Application Industry Technology Alliance, told the Global Times. "Chinese models may not have the highest performance ranking, but for many people they can do the job at a much cheaper cost."

DeepSeek's increased revenue-generating capability is accompanied by broader revenue growth among leading Chinese AI firms amid a broader commercialization push and global competition, according to analysts and reports.

Chinese AI developer Z.ai recently raised its year-end annual recurring revenue outlook by 25 percent to $3 billion, up from a previous estimate of $2.4 billion. MiniMax, another Chinese AI firm, reported in August a 283-percent surge in first-half revenue to $116.6 million. In 2025, the company's full-year revenue was $79 million.

Behind the surging revenue of Chinese AI firms is growing global use of their models. Chinese LLMs have outpaced their counterparts from the US in weekly usage for 22 consecutive weeks, firmly holding the top spot globally, according to OpenRouter. The platform's latest data showed that Chinese AI models processed 62.22 trillion tokens from September 21 to 27 compared with 14.2 trillion generated by US models. 

Rising revenue at Chinese AI firms has demonstrated the growing competitiveness of Chinese LLMs in performance, service quality and overall usage costs as a broader commercialization push takes hold, Chinese experts said.

"From Z.ai and MiniMax to DeepSeek and others, the API revenues of China's first-tier players are rising rapidly in tandem, showing that this is an industry-wide shift, not a coincidence at individual companies," Tian said, noting that Chinese LLM companies have shifted gears.

"Chinese large model firms are moving from competing on low prices to value-based pricing, and the role of open source has shifted from a source of free supply to a customer acquisition channel," Tian said. "Most importantly, Chinese AI firms are now able to generate revenue from their products and services instead of relying on investor cash-burning."

Strong potential 

Another key topic in focus is the comparison between Chinese and US leading AI firms' revenue-generating capabilities.

Chinese experts noted that API usage and user base are currently two of the most important factors determining AI companies' revenue-generating capabilities. However, there are significant differences in pricing and demand between leading Chinese and US AI firms, as companies pursuing open-weight models and those developing top-tier closed systems follow different approaches to generating revenue.

After its recent price hike, DeepSeek-V4 Pro was priced at $3.96 per million output tokens, according to a price list on its official website. This compares with the $20 per million output tokens charged for Claude Opus 5.5 by Anthropic. 

This could also explain why the latest annualized revenue figure for DeepSeek trails OpenAI's reported revenue run rate of $40 billion and Anthropic's reported $65 billion by the end of July.

However, Tian said it is worth noting that investors are willing to accept DeepSeek's market value at 75 times its annualized revenue, while they are paying only 15 times for Anthropic, whose market valuation reportedly topped $965 billion in May. That difference is important, Tian said, noting that investors value the firm's ability to develop frontier models with far fewer resources.

"AI development in China and the US exhibits distinct characteristics: While US tech giants dominate benchmark rankings, Chinese companies deliver more affordable models to customers and are expanding rapidly, as proliferation of AI applications domestically fuels growing demand for their services," Xiang said.

As Chinese model firms' pricing is only a fraction of that of US firms and Chinese clients' habit of paying for inference is still maturing, Chinese AI firms have strong potential to grow further thanks to their large global user base and rising demand for coding agents, analysts said. 

Amid increasing real-world AI application scenarios in China, tens of millions of end devices are being integrated with LLMs, driving up demand for API calls, which is a boon for companies like DeepSeek, Ma Jihua, a veteran tech industry analyst, told the Global Times.

With several leading Chinese AI services boasting active user bases of over 100 million, AI has entered the daily lives of Chinese people, providing a strong foundation for AI companies to generate revenue growth, said Ma.

Notably, China's generative AI user base had exceeded 700 million by the first half of 2026, with the adoption rate above 50 percent, according to the Generative Artificial Intelligence Application Development Report (2026) released by an institute under the China Internet Network Information Center.

Chinese AI companies also enjoy the advantage of better cooperation and coordination among companies in China. This contrasts with the situation in the US, where different tech companies often compete for exclusive market dominance for their own models, said Xiang.

Amid increasingly diversified AI application scenarios in the Chinese market across sectors ranging from industrial production, robots, drones, intelligent connected vehicles, entertainment and even toys, demand for large models has steadily expanded.

These trends contribute to Chinese LLM firms' revenue, which are moving along a track of inclusive development aimed at making AI affordable to more companies and individuals, Ma said. 

"With volume comes revenue. Chinese LLM firms' prospects are bright," Ma said.