Short-selling surges in Zhipu AI vs MiniMax Ahead of Earnings
Short-selling activity has surged to record highs against Hong Kong-listed AI firms Zhipu AI (Z.AI Co.) and MiniMax Group ahead of upcoming earnings. Short-selling volumes peaked in June 2026, highlighting investor anxiety in China’s generative AI price war.
Zhipu has outperformed sharply since late March. Its stock rallied about 170% through mid-June 2026, while MiniMax fell nearly 50% over the same period. This divergence is driving a pair trade strategy: investors go long Zhipu and short MiniMax to profit from widening performance gaps.
The data shows the intensity of short-selling: Zhipu’s shorts reached 178,900 shares (about HK$430 million), while MiniMax’s hit 239,700 shares (around HK$141 million). Despite fewer shares, Zhipu’s short position is larger in value, reflecting Zhipu’s higher share price.
The market appears to be picking sides on technology and pricing power. Zhipu’s GLM-5.2 is viewed as having an edge over MiniMax’s M3, especially for enterprise use cases where customers pay for access. Neither company is profitable, and both remain heavy on R&D.
A structural factor may intensify pressure on MiniMax: July 2026 lock-up expirations could free about 65% of MiniMax’s shares for trading, versus roughly 6% for Zhipu. That can reduce borrowing costs and amplify selling. Bernstein started coverage on Aug. 4, 2026: Zhipu rated Outperform (target HK$1,350) and MiniMax rated Market-Perform (target HK$275).
Neutral
This news is bearish for MiniMax specifically and bullish for Zhipu, but its direct link to the crypto market is limited. For crypto traders, the main relevance is risk sentiment and volatility: record short-selling into earnings can spill over into broader “tech/AI risk-off” positioning, which occasionally tightens liquidity across high-beta assets (including some crypto segments).
In the short term, the lock-up dynamic (65% of MiniMax shares becoming borrowable vs ~6% for Zhipu) can increase downside pressure and amplify volatility around earnings. That can make traders more cautious, sometimes leading to reduced appetite for speculative trades. In the longer term, the market’s focus on model quality (GLM-5.2 vs M3) and pricing power reinforces a “winner-takes-more” pattern seen in prior tech cycles: flows concentrate toward firms perceived to have durable enterprise revenue.
However, because the catalyst is equity-specific (HK-listed AI firms) rather than a macro or crypto-structure event, systemic impact on crypto stability is likely limited. Net effect for crypto markets: neutral, with potential short-term sentiment-driven volatility rather than a sustained directional impulse.