China's AI Landscape: Token Economy and Market Trends

Explore the current state of AI in China, including the rise of the token economy, major players, and market dynamics.

Introduction

On April 12, Fortune magazine published an article titled “The Rise of Intelligent Agents and the Naming of ‘Tokens’: An Overview of China’s AI Landscape” by Nicholas Gordon. This article summarizes the current state of AI in China.

Token Naming and Usage

China has officially named the basic unit of information processing in large language models as “Token” (词元). Liu Liehong, head of the National Bureau of Statistics, announced this name last month, explaining that tokens are now the “settlement unit connecting technology supply and business demand.” In March, the bureau revealed that the daily usage of tokens in China has surged from 100 billion at the beginning of 2024 to over 140 trillion. Data from the global AI model aggregation platform “Open Router” indicates that the weekly usage of AI large models in China has surpassed that of the United States.

Competitive Advantage

Mohit Kumar, a global macro strategist at Jefferies Group, stated, “We believe China will emerge as the big winner in this tech race for several reasons: valuation advantages, higher AI application penetration, and power generation advantages.”

China’s current goal is to leverage a large number of efficient open-source large models to build a “token economy” and promote the application of AI in real-world scenarios.

Major Players in the Market

E-commerce giant Alibaba is investing in open-source large models that developers can download and modify for free. Due to low usage costs, Qianwen has become an ideal choice for many startups. Qianwen has gained popularity among developers in Southeast Asia and the Middle East, as well as recognition from Western users. The Muse Spark model, recently launched by the American Metaverse Super Intelligence Laboratory, partially drew on Qianwen during its training process.

In contrast to Alibaba, ByteDance’s AI models are mostly closed-source, capturing the market with advantages in product design and user experience. The company’s “Doubao” is currently one of the most widely used AI applications in China.

Tencent launched the WeChat ClawBot plugin last month, supporting access to the “Open Claw” intelligent agent project, allowing users to interact with OpenClaw directly through WeChat chats.

Additionally, large tech companies like Xiaomi and Meituan have also introduced their self-developed large models, which are expected to reshape the industry landscape.

Recognition from Silicon Valley

A new generation of Chinese AI startups is gaining recognition in Silicon Valley. When the American code editor “Cursor” released Composer 2, observant users noticed that the model was developed based on the open-source model Kimi K2.5 from Beijing’s Dark Side Technology Co., Ltd.

Shanghai Xiyu Technology Co., Ltd. (MiniMax) is projected to achieve a total revenue of $79 million in 2025, a 159% year-on-year increase, with 70% of its revenue coming from overseas markets, indicating a growing global demand for Chinese large models.

Rapid Development in Physical AI

With advantages in low-cost supply chains, China is also experiencing rapid growth in the field of “physical AI.” Notable humanoid robot startup Yushu Technology has applied for an initial public offering on the Shanghai Stock Exchange’s STAR Market. Zhiyuan Robotics and UBTECH are also leading robot startups in China.

Chinese companies are accelerating their entry into the autonomous driving sector. Earlier this month, Pony.ai partnered with Croatia’s Verne company and ride-hailing platform Uber to launch a commercial autonomous taxi service in Zagreb, becoming the first public-facing commercial service in Europe. WeRide has also partnered with Uber to launch a fully commercial autonomous taxi service in Dubai.

AI Acceptance Among Users

Chinese users show a far higher acceptance of AI compared to Western users. A survey by Edelman in October last year revealed that 87% of Chinese respondents trust AI, while the figure in the U.S. was only 32%.

The booming short drama industry in China reflects consumer acceptance of AI. With AI tools, the cost of producing a short drama is only one-tenth of traditional methods, and the production cycle has been shortened from 15-30 days to under 5 days.

Policy Advantages

China’s most significant policy advantage may not be directly related to artificial intelligence. In recent years, China has actively enhanced its power generation and transmission capabilities. Goldman Sachs predicts that by 2030, China will have approximately 400 gigawatts of backup power capacity, about three times the expected electricity demand of global data centers.

Challenges Ahead

However, both the Chinese and American AI industries face a common challenge: how to convert tokens into actual profits. Profit pressures have prompted some Chinese tech companies to adjust their strategies. Additionally, several companies, including Baidu, have raised prices for models and cloud services.

Was this helpful?

Likes and saves are stored in your browser on this device only (local storage) and are not uploaded to our servers.

Comments

Discussion is powered by Giscus (GitHub Discussions). Add repo, repoID, category, and categoryID under [params.comments.giscus] in hugo.toml using the values from the Giscus setup tool.