According to Fortune, while Silicon Valley remains captivated by flashy demos, Beijing has officially named the currency of its digital economy: Ciyuan.
With processing speeds hitting 140 trillion tokens per day and a wave of AI-driven companies going public, China is no longer playing catch-up. Instead, it is forging a distinct ecosystem—practical, ambitious, and fast-moving.
The “Ciyuan” Revolution
For the first time, China has introduced an official term for “tokens” in large language models: Ciyuan. Liu Liehong, head of the National Data Administration, unveiled the term in March, describing it as a “payment unit connecting technological supply with commercial demand.”
The numbers illustrate the surge. China now processes around 140 trillion ciyuan daily, up from just 100 billion at the start of 2024. On OpenRouter, a popular AI marketplace, Chinese code has begun to outperform U.S. counterparts in efficiency and popularity.
AI has also become a lifeline for Chinese tech giants emerging from years of regulatory crackdowns. In Hong Kong, IPO activity has hit a five-year high, fueled by unicorns like MiniMax, Zhipu AI, and chipmaker Biren.
“China is the big winner in this tech race for several reasons: attractive valuations, widespread AI adoption, and energy production advantages,” said Mohit Kumar, global macro strategist at Jefferies, during the bank’s Asia Forum in March.
Consumer Confidence and Industry Growth
Public sentiment is a major driver. An Edelman survey last October found that 87% of Chinese citizens trust AI, compared to just 32% in the U.S. This openness has accelerated the rise of “practical AI” industries.
One standout example is short films: around 470 new productions debut daily, with AI slashing costs from 1 million yuan to 100,000 yuan ($14,600) and reducing production time from weeks to under five days.
Tech Giants Take Different Paths
Alibaba is betting on open-source. Its Qwen model allows developers to freely download and customize, attracting users from Southeast Asia to the Middle East. Meta’s latest Muse Spark model was even partially trained on Qwen. CEO Eddie Wu recently reorganized Alibaba’s AI operations into the “Alibaba Token Hub” (ATH), declaring: “ATH’s mission is to create, distribute, and apply tokens.”
ByteDance, by contrast, has opted for a closed ecosystem focused on user experience. Its Doubao chatbot is now China’s most-used AI app, boasting 100 million daily active users during Lunar New Year. Tencent has also entered the race with ClawBot, integrated directly into WeChat, giving over a billion monthly users access to AI within their messaging app.
Hardware AI and Profitability
Beyond software, Chinese AI startups in hardware are leveraging supply chain advantages. Unitree Robotics filed for a 4.2 billion yuan ($610 million) IPO on Shanghai’s STAR Market, reporting adjusted net profits of 600 million yuan ($87 million)—a rarity in robotics.
Meanwhile, pure AI labs remain unprofitable. MiniMax posted $79 million in revenue in 2025 but lost $250 million. Zhipu AI earned $104 million yet reported losses of $680 million due to soaring R&D costs. Still, investor enthusiasm is strong: Zhipu’s stock has surged over 570% since its IPO, while Moonshot AI—backed by Alibaba and HongShan—has reached a $18 billion valuation despite Western caution over U.S. chip sanctions.
Challenges Ahead
Chinese firms still face hurdles from U.S. export controls on advanced chips. Alibaba has responded by launching a data center powered entirely by its self-designed Zhenwu chips, though performance remains uncertain compared to American supply chains.
Internal tensions are also emerging. Manus AI, for example, quietly relocated to Singapore before being acquired by Meta for $2 billion. Its founders now face exit bans imposed by Beijing.
Conclusion
China is rapidly constructing a “token economy” built on efficiency and real-world application. While capital expenditures by giants like Alibaba ($17 billion) and ByteDance ($23 billion projected) remain below those of Alphabet and Meta, China’s energy surplus (expected to reach 400 gigawatts by 2030) and overwhelming domestic support give its AI industry strategic advantages the West cannot ignore. |