Hong Kong AI stocks plunge: Zhipu loses 20%

Hong Kong's AI stock market was swept by a wave of selling: Zhipu AI's shares collapsed by more than 20%, MiniMax shares lost around 13%. In a single trading day, the entire segment of companies specializing in developing large language models came under pressure — and this decline says much more than mere exchange statistics.

To understand the scale of what happened, it is important to recall the backdrop against which the history of Chinese AI is unfolding. Just a few months ago, investors enthusiastically bought shares of local language model developers, betting that China would form its own ecosystem, independent from Western platforms. Zhipu AI — one of the market pioneers, standing behind the GLM model — was considered a symbol of these expectations. MiniMax, which developed multimodal solutions and attracted an audience of hundreds of millions of users, was perceived as proof that competitive products in China are possible. Both players attracted serious venture capital and enjoyed the reputation of technological flagships.

What changed? The fall in share prices became a concentrated expression of several concerns that had been building in the investment community. First, the global race in the field of large language models sharply accelerated: the emergence of DeepSeek with models demonstrating impressive performance-to-training-cost ratios changed conventional benchmarks.

It suddenly became clear that the expensive path of scaling computational capacity — not the only route to competitive advantage. Companies whose business models were built on the assumption of continuous infrastructure expansion faced a reassessment of their market value. Second, the question of monetization remains open: between the ability to create impressive demonstrations of model capabilities and the ability to convert them into sustainable revenue — there is an enormous gap that most Chinese LLM startups have yet to overcome.

The selloff also reflects the structural vulnerability of the entire segment. Investors acquired shares of companies operating at early stages of monetization, betting that the market for corporate AI solutions in China would explode with growth in the next year or two. However, the pace of adoption turned out to be slower than predicted: large state-owned enterprises are cautious in making decisions, and small and medium-sized businesses are not yet ready to pay premium prices for tools whose value requires proof in concrete industry scenarios. Add to this geopolitical pressure — restrictions on chip exports, complicating access to advanced GPUs — and the picture becomes even more ambiguous.

For the broader technology industry, today's movement in quotes serves as a reminder of the gap between narrative and reality. The excitement around large language models created an environment in which any announcement of a new model or partnership was perceived by the market as a buy signal. Now the pendulum has swung the other way: investors are beginning to ask tough questions about profitability, product differentiation, and the company's ability to survive in the price war that more capitalized players can unleash — whether it is Alibaba with Qwen, Baidu with Ernie, or international competitors whose models are becoming increasingly accessible.

Neither Zhipu AI nor MiniMax will disappear after one trading day. Both companies possess real technologies, teams, and partnership portfolios. But the stock market correction poses an uncomfortable question that can no longer be postponed: what exactly constitutes their indispensability to a corporate or consumer client who will be choosing between a dozen comparably quality solutions in a year? The answer to this question will determine whether the current decline will turn out to be temporary turbulence or the beginning of a painful sobering for the entire market of Chinese LLM companies. For now, the market sends an unambiguous signal: the time when simply being in the large language model niche was enough is ending.