Narrative-driven everything, China's new AI giants enter an era of extreme volatility, retail investors flooding in!

Can AI · Token Narratives Mask the Reality of Huge Losses in AI Companies?

Emerging Chinese AI stocks have become one of the most volatile sectors in Asian stock markets. Driven by fervor around AI narratives, retail investors dominate pricing, with institutional ownership remaining low. This combination is pushing related stocks into extreme volatility ranges.

Measured by 90-day annualized volatility, half of the top ten most volatile stocks among Asian companies with market caps over $1 billion are recent Chinese AI startups.

Moore Threads surged over 700% in five trading days before nearly halving; AI large model developer MiniMax Group Inc. has gained over 500% since its January listing.

As some companies push to be included in the Shanghai-Shenzhen-Hong Kong Connect trading mechanisms, mainland investors’ access channels are about to open. Given mainland investors’ natural preference for momentum trading, the volatility of related stocks may intensify further, posing significant risks for investors trying to chase gains.

Retail-led Pricing, Institutional Holdings Extremely Thin

Chinese stocks, especially those listed in Hong Kong, have long experienced speculative volatility, but the recent AI stock turbulence has structural roots: institutional investors are almost absent.

Exchange data shows that only about 9.3% of MiniMax’s shares are held by disclosed institutional investors; the rest are owned by individuals and investors not required to report holdings.

Among the five most volatile stocks in Asia, the average institutional shareholding is only 13%. In contrast, Tencent Holdings and Alibaba Group have institutional ownership around 50%.

The weak institutional base means stock prices are driven almost entirely by sentiment and capital flows, not fundamentals. In March, MiniMax’s intraday volatility averaged 14 percentage points; Zhipu’s intraday high-low spread reached 13 percentage points, while Alibaba’s was only 3.6%.

Token Demand Boom Ignites Market, Performance Losses Remain a Concern

The core narrative fueling this round of speculation is market bets on a surge in “token” demand. Tokens measure the amount of data units processed by AI models and are the fundamental metric for outputs like text generation. Investors are beginning to believe that providers like MiniMax and Zhipu will benefit most directly from this demand wave.

This expectation is not entirely without fundamentals. According to Bloomberg Intelligence, revenue growth for MiniMax and Zhipu is forecasted to exceed 150% year-over-year by 2028, benefiting from the direct monetization of token usage.

MiniMax’s latest AI product, MiniMax M2.7, ranks third in the popular model list tracked by OpenRouter, surpassing models from DeepSeek and Alibaba.

However, behind these impressive product performances are heavy losses. MiniMax posted a net loss of $1.9 billion last year; Zhipu’s net loss is projected to widen by 60% in 2025. Despite this, Zhipu’s stock surged 35% intraday on Wednesday, and MiniMax rose 13%. Investors are ignoring losses, betting on the long-term realization of token demand and pricing power.

“Unproven Business Models,” Experts Warn of Chasing Risks

Market professionals remain cautious in the face of this hype.

“All these new AI stocks’ valuations are almost entirely based on their AI roadmaps,” said Jasmine Duan, Senior Investment Strategist at RBC Wealth Management Asia.

“Chasing gains carries risks because some companies’ business models are unproven, and changes can be quite rapid.”

Luo Jing, Investment Director at Value Partners Group, pointed out that computing infrastructure remains a decisive factor in China’s AI landscape, with established industry leaders maintaining strategic advantages in long-term competition.

Daniel So, Senior Trading Strategist at Goldhorse Capital Management, noted that given the uncertain long-term outlook, his clients prefer shorter holding periods when dealing with companies like MiniMax and Zhipu. He said:

“They are unprofitable, valuation is hard to quantify, high potential but low certainty—volatility will persist.”