Bernstein has started coverage of China's listed artificial intelligence laboratories, presenting a broadly optimistic long-term outlook for the industry and issuing divergent ratings for two prominent names. The brokerage opened Outperform coverage on Z.ai and Market-Perform on Minimax, while forecasting that continued frontier AI work will remain an important technological and strategic focus in China.
The research note highlights several structural drivers the brokerage sees supporting growth in the sector. Bernstein argues that easing domestic computing constraints, advances in model reasoning and wider adoption of lower-cost open-source models should help leading Chinese AI developers scale commercial applications. The firm estimates China's AI market could yield between $100 billion and $200 billion in annual revenue, excluding consumer applications, and says ongoing investment in domestic AI infrastructure should underpin sustained expansion.
On Z.ai, Bernstein identifies the company as its preferred pick and assigns a HK$1,350 price target. The brokerage points to Z.ai's research pedigree and its competitive GLM-5.2 model, and it expects the company's revenue to outpace current consensus forecasts in the coming years. Bernstein also projects that Z.ai may reach non-GAAP operating breakeven around 2028 despite ongoing research and development spending.
Bernstein notes that recent weakness in Z.ai's share price after the launch of rival Kimi K3 has lowered market expectations, but maintains that Z.ai remains well positioned. The firm emphasizes that Z.ai's coding-focused AI models continue to rank among China's most competitive and flags upcoming releases - including GLM-5.3 and next-generation pre-trained models - as potential catalysts for the stock.
By contrast, Minimax receives a Market-Perform rating with a HK$275 price target. Bernstein characterizes the company's next-generation M3 Pro model as a "make-or-break" release following what it describes as the underwhelming performance of M3. While the brokerage expects annual recurring revenue growth to improve as newer models roll out, it views Minimax's strategic focus on AI video generation as offering a smaller long-term revenue opportunity versus coding and agentic AI applications.
More broadly, Bernstein anticipates the nature of competition among AI labs in China to shift over time. As more tasks become commercially viable, the brokerage expects emphasis to move from raw reasoning capability toward cost efficiency and compute availability, even as research capability and frontier model performance remain primary valuation drivers for AI labs.
Key points
- Bernstein initiates coverage on China's listed AI labs, favoring Z.ai with an Outperform rating and Market-Perform on Minimax.
- The brokerage projects China’s non-consumer AI market could generate $100 billion to $200 billion annually, supported by domestic infrastructure investment.
- Competition is expected to evolve from pure reasoning performance to a greater focus on cost-efficiency and compute availability as commercial applications scale.
Risks and uncertainties
- Z.ai: near-term share-price volatility and model release execution risks could affect the outlook despite Bernstein's positive view.
- Minimax: the success of the M3 Pro model is framed as a make-or-break event after M3's underperformance, introducing product execution risk.
- Sector-wide: shifts in compute availability and cost structures could alter competitive dynamics and revenue trajectories for AI developers.