China’s AI Market Thrives Amid Global Enthusiasm

China’s affordable artificial intelligence models are gaining significant traction worldwide, bringing new players to the forefront of its stock market.

As AI evolves to handle a variety of tasks, a burgeoning economy focused on “tokens”—the fundamental units of data in large-scale language models—has emerged. Increased global token usage has positioned China advantageously, bolstered by low-cost electricity and a plethora of AI model developers.

Niche startups are attracting more investment than established tech giants, driven by the promising growth potential of AI. Recent entrants like MiniMax Group Inc. and Knowledge Atlas Technology Joint Stock Co., known as Zhipu, have already achieved market valuations exceeding $40 billion, surpassing well-known names like Baidu Inc. and Kuaishou Technology.

Even major firms, such as Alibaba Group Holding Ltd., have seen substantial stock increases despite financial setbacks. Investors like Victoria Mio—who manages portfolios at Janus Henderson Group—believe significant changes are unfolding in the market.

“This rally denotes a fundamental revaluation, not merely a fleeting trade,” Mio stated. “Investors are noticing tangible token consumption and early pricing power, indicating that AI inference is transitioning into a revenue-generating enterprise, rather than just being a buzzword.”

Much like the previous year’s tech boom, investors have a growing consensus that China will emerge as a long-term winner due to its cost efficiencies. The Chinese government is leveraging this momentum to enhance the global competitive stance of its tech industry, with “token exports” becoming quite the topic in state media.

Understanding Tokens

A token is essentially a data chunk that represents roughly four characters. AI firms typically offer free and fixed monthly service plans, but users, especially those developing applications, often surpass these limits and pay for additional token usage. This includes fees for both the input and output tokens, with the latter being significantly more expensive.

Tokens have now become the cornerstone of the burgeoning AI economy, facilitating transactions not just between businesses and users, but also serving as a metric for coding productivity within tech companies.

According to LLM distribution platform OpenRouter, the top three models for token consumption this month were all from Chinese firms: Xiaomi Corp.’s MiMo, Alibaba’s Qwen, and DeepSeek. Notably, MiniMax has two models listed in the top ten, along with another from the startup StepFun.

“China has a competitive edge in providing tokens and models globally,” said Bush Chew, an investment manager at Aberdeen Standard Investment Hong Kong Ltd. “Constructing a data center in China is considerably less expensive compared to the US.”

Remarkably, as of April 19, MiniMax ranks fourth in market share, only behind American giants like Alphabet Inc., Anthropic PBC, and OpenAI. The most sought-after models from MiniMax charge about $1 for a million output tokens, while certain Google models can reach $3, and Anthropic’s offerings go beyond $15.

Goldman Sachs views MiniMax as a top contender among Chinese AI firms with the potential for significant growth, attributing this to comprehensive multimodal products and cost advantages in token usage.

MiniMax and Zhipu are part of a select group known as the “Six Tigers,” startups that offer specialized AI services and often have backing from larger corporations, like Alibaba’s stake in MiniMax.

Both companies went public in Hong Kong this January, with Zhipu shares skyrocketing about 670% since then, while MiniMax shares have jumped nearly 440%. In contrast, Alibaba’s stock has declined by 5%, and Tencent Holdings has seen a 15% drop.

Emerging Players and Market Trends

As investors flock to new startups, more firms such as StepFun, Moonshot AI, and Baichuan are expected to announce IPOs. The soaring share prices of fledgling companies that still operate at a loss have raised some concerns among investors.

“The valuation surge in some instances has occurred rapidly, likely outpacing actual earnings growth,” remarked Nicholas Chui, a fund manager at Franklin Templeton. “The introduction of robust models will be vital moving ahead. Continued adaptation by companies and individuals both in Japan and internationally could drive support.”

The market remains a competitive arena, with new entrants challenging even well-established firms. Tencent, for example, could leverage its popular WeChat app for greater advantage, although recent efforts have not yet met expectations.

Cost remains a crucial consideration for users. The high demand for tokens associated with OpenClaw—an application that automates everyday tasks—illustrates how affordable models are appealing to tech consumers. Additionally, a trend dubbed “token maximization,” where engineers optimize token utilization to showcase productivity, is gaining momentum.

While the United States still holds a technological edge, China’s low-cost AI solutions are effectively capturing significant global demand. The latest figures show that China’s average daily token consumption topped 140 trillion in March, a considerable jump from 100 billion at the start of 2024.

“For investors, China’s AI landscape should be analyzed through the lens of usage rates, cost efficiencies, and monetization capabilities rather than merely the performance of leading models,” suggested Mio. “China is effectively industrializing AI services at minimal marginal costs and seamlessly integrating them into practical workflows.”