Felix Wang calls the Chinese AI stock 'one of the biggest bubbles I've ever seen' and sets a fair value target nearly 70% below current trading levels.
Zhipu has been the best-performing stock on the Hang Seng Tech Index this year. One veteran analyst thinks that’s exactly the problem.
Felix Wang of Hedgeye Risk Management is recommending a short position on the Chinese AI developer, pegging its fair value at HK$407. That’s roughly 70% below where shares currently trade, in the range of HK$1,400 to HK$1,500.
From IPO darling to bubble territory #
Zhipu, the company behind the GLM series of large language models, went public on January 8 at approximately HK$116 per share. In less than six months, the stock surged more than 1,100%, hitting intraday peaks above HK$1,600 in mid-June.
Wang described Zhipu as “one of the biggest bubbles I’ve ever seen,” citing eroded pricing power in the wake of DeepSeek’s aggressive moves.
DeepSeek’s price war changes the math #
In May, DeepSeek finalized a permanent 75% price cut on its V4-Pro model. The result: API costs dropped to approximately $0.0035 per million input tokens.
Wang’s core argument is that Zhipu lacks the pricing power to compete in this environment. When the commodity you’re selling, in this case large language model API access, is being aggressively discounted by well-funded rivals, revenue projections built on higher price assumptions start to crumble.
Index inclusion provided a temporary lifeline #
Zhipu did catch a break in June when it was added to the Hang Seng Tech Index alongside competitor MiniMax. Index inclusion typically drives passive buying from funds that track the benchmark, providing a mechanical boost to share prices regardless of fundamentals.
The company also open-sourced its GLM-5.2 model, a move designed to expand its developer ecosystem. Both developments helped stabilize the stock after a period of weakness in April, when competitive pressures from DeepSeek first started weighing on sentiment.
Index inclusion is a one-time event. Once the passive flows are absorbed, the stock has to stand on its own merits. And open-sourcing a model, while strategically sound for ecosystem building, doesn’t directly solve the revenue problem that Wang is flagging.
What this means for investors #
Wang’s HK$407 fair value estimate implies that Zhipu would need to fall roughly 70% from current levels to reach what he considers reasonable.
For traders with exposure to Zhipu or other Chinese AI stocks, the key variable to watch is pricing trends across the sector. If more competitors follow DeepSeek’s lead and slash API costs, the entire valuation framework for companies like Zhipu needs to be rebuilt from scratch. Zhipu now needs to deliver results that justify a valuation more than ten times its IPO price, and it needs to do so while its products are being commoditized in real time.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our