Record short bets pile up against Zhipu AI and MiniMax Group as China’s AI rivalry intensifies

Short-selling volumes for Zhipu AI (Z.AI Co.) and MiniMax Group Inc. hit all-time highs in June 2026, with Zhipu recording 178,900 shares shorted (HK$430 million) and MiniMax 239,700 shares (HK$141 million), as investors bet on divergent fortunes amid China's generative AI price war. Zhipu's stock surged 170% through mid-June, while MiniMax fell 50%, and Bernstein on August 4, 2026 rated Zhipu Outperform with a HK$1,350 target and MiniMax Market-Perform with a HK$275 target. Lock-up expirations in July 2026 freed 65% of MiniMax's shares versus 6% for Zhipu, amplifying short pressure on MiniMax.

Photo: Tinny HU / Pexels Record short bets pile up against Zhipu AI and MiniMax Group as China’s AI rivalry intensifies Short-selling volumes for both Hong Kong-listed AI firms hit all-time highs ahead of earnings, with a pair trade strategy emerging among investors betting on divergent fortunes. Short-selling activity against two of China’s most prominent AI companies has surged to unprecedented levels, reflecting deep investor anxiety about which players will survive the country’s increasingly brutal generative AI price war. Zhipu AI Z.AI Co. and MiniMax Group Inc., both listed on the Hong Kong stock exchange in early 2026, have seen their short volumes climb to record highs ahead of upcoming earnings reports. The bet isn’t just that AI stocks are overheated. It’s that one of these two companies is positioned far better than the other. A tale of two AI stocks The performance gap between Zhipu and MiniMax since late March has been striking. Zhipu’s stock surged roughly 170% through mid-June 2026, while MiniMax’s shares cratered nearly 50% over the same window. That kind of divergence in a sector where both companies are pursuing essentially the same market tends to attract a specific breed of investor: the pair trader. The emerging strategy is straightforward. Go long on Zhipu, short MiniMax, and pocket the spread as the gap widens. Short-selling volumes peaked in June 2026 at levels neither stock had previously seen. Zhipu recorded short volumes of 178,900 shares, valued at approximately HK$430 million. MiniMax hit 239,700 shares, worth around HK$141 million. Both numbers represent all-time highs. The fact that Zhipu’s short volume carries a higher dollar value despite fewer shares reflects its substantially higher stock price, a direct consequence of the 170% rally that has made it a magnet for both bulls and bears simultaneously. Why the market is picking sides The core thesis driving the divergence comes down to technology and pricing power. Zhipu’s GLM-5.2 model is widely perceived to hold a meaningful edge over MiniMax’s M3 model, particularly when it comes to enterprise applications where customers actually pay for access. Neither company is profitable. Both carry substantial R&D expenditures, a necessity in a sector where falling behind technologically can be fatal to a business model. Bernstein initiated coverage of both stocks on August 4, 2026, making the contrast explicit. Zhipu received an Outperform rating with a target price of HK$1,350. MiniMax got a Market-Perform rating with a target of HK$275. The lock-up effect A structural factor is amplifying the pressure on MiniMax. Lock-up expirations in July 2026 freed approximately 65% of MiniMax’s outstanding shares for trading, dramatically increasing the supply available for short sellers to borrow. Zhipu, by contrast, saw only about 6% of its shares become tradable during the same period. That asymmetry creates a mechanical disadvantage for MiniMax. More borrowable shares means lower borrowing costs for short sellers, which means more shorts, which means more downward pressure on the stock. Both companies secured funding from Alibaba and Tencent before going public, giving them the kind of strategic backing that typically signals staying power. But backing from tech giants hasn’t insulated MiniMax from the market’s verdict on its competitive positioning. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .