Z.AI vs MiniMax: Which Chinese frontier AI stock should you buy?
Z.AI vs MiniMax: Which Chinese frontier AI stock should you buy?
Two Chinese AI stocks, one radically different tale: Z.AI has surged +767.5% YTD against MiniMax’s more modest +7.8% — yet both sit well below their 52-week peaks, raising the key question: which has more runway? Wall Street’s freshest answer landed just this week, with Bernstein firmly backing Z.AI as the "cleaner expression" of frontier AI R&D.
The Numbers, Side by Side
| Metric | Z.AI (2513.HK) | MiniMax (0100.HK) |
|---|---|---|
| Price | HK$1,041 | HK$253.80 |
| 52-Week Range | HK$116.10 – HK$2,980 | HK$186.20 – HK$1,330 |
| YTD Return | +767.5% | +7.8% |
| 6-Month | +381.5% | -47.8% |
| 3-Month | +13.2% | -67.6% |
| 1-Month | -41.9% | -26.8% |
| Market Cap | HK$464.1B | HK$79.6B |
| Bernstein Target | HK$1,350 (Outperform) | HK$275 (Market-Perform) |
| JPMorgan Target | — | HK$160 (Neutral) |
| Implied Upside | ~+30% | ~+8% / -37% (JPM) |
Z.AI’s Coding Edge
Z.AI (Zhipu) is advancing on multiple fronts. Shares rose nearly +20% on Jul 31 after launching paid subscriptions for its GLM coding plan — a direct monetization signal that the market has been watching for. Bernstein, initiating with Outperform and a HK$1,350 target, views it as a dominant player in coding and agentic AI — the segments with the largest long-term revenue addressable market. The upcoming GLM-5.3 and next-gen models are cited as near-term catalysts, with non-GAAP operating breakeven targeted around 2028.
The caveat? Z.AI is already ~6x larger by market cap than MiniMax and trades 65% off its 52-week high of HK$2,980. After a 768% run, the margin of safety is thin and a -41.9% pullback in just one month points to significant volatility.
MiniMax’s Video Gamble
MiniMax is betting heavily on multimodal video AI. Its new H3 model — generating 2K video at under one-third the cost of rivals — drove a +17.4% gain on Jul 31. The open-weight strategy could build developer adoption rapidly.
But the bear case is pointed: Bernstein views multimodal video as having a lower revenue ceiling than coding/agentic AI, and its upcoming M3 Pro model is described as "make-or-break." JPMorgan’s price target cut to HK$160 — implying -37% downside from current levels — reflects genuine skepticism about the model lineup’s competitiveness. Competition is also intensifying, with Alibaba’s Qwen3.8-Max and DeepSeek’s V4-Flash all vying for the same market.
The Verdict
Analyst consensus tilts clearly toward Z.AI. Its coding + agentic AI focus sits in the larger, more defensible part of China’s AI stack. Bernstein’s ~30% upside to HK$1,350 is meaningful given the recent correction. MiniMax offers higher volatility upside if M3 Pro surprises, but the risk/reward is asymmetric — JPMorgan’s HK$160 target is a sobering floor.
Insights
- Z.AI’s 6M return of +381% vs MiniMax’s -48% reflects a fundamental divergence in model competitive positioning, not just sentiment
- Both stocks are deeply below 52-week highs — Z.AI off 65%, MiniMax off 81% — the peak euphoria phase has passed for both
- China AI stocks face a shared macro headwind: US tech export restrictions and intensifying domestic competition from DeepSeek and Qwen