Bernstein starts coverage of China’s AI labs, favors Z.ai over Minimax

Investing.com -- Bernstein initiated coverage of China's listed artificial intelligence labs with an optimistic long-term outlook for the sector, launching Outperform coverage on Z.ai and Market-Perform on Minimax as it expects frontier AI development to remain a key technological and strategic priority in China.

The brokerage argued that China's leading AI developers are well positioned to benefit from easing domestic computing constraints, improving reasoning capabilities and expanding adoption of lower-cost open-source models. It estimates China's AI market could generate $100 billion to $200 billion in annual revenue, excluding consumer applications, while continued investment in domestic AI infrastructure should support long-term growth.

[[pro_promotion |

Bernstein said Z.ai is its preferred pick, assigning a HK$1,350 price target, citing the company's research pedigree, competitive GLM-5.2 model and expectations that revenue will exceed consensus estimates in coming years. The brokerage expects the company to achieve non-GAAP operating breakeven around 2028 despite continued investment in research and development.

The firm said recent share price weakness following rival Kimi K3's launch had reset expectations, but maintained that Z.ai remains well positioned as its coding-focused AI models continue to rank among China's most competitive. It expects upcoming GLM-5.3 and next-generation pre-trained models to be important catalysts for the stock.

By contrast, Bernstein assigned Minimax a Market-Perform rating with a HK$275 price target, arguing that its next-generation M3 Pro model has become a "make-or-break" release after the underwhelming performance of M3. While the brokerage expects ARR growth to improve with newer models, it believes the company's focus on AI video generation offers a smaller long-term revenue opportunity than coding and agentic AI applications.

More broadly, Bernstein expects AI competition to increasingly shift from raw reasoning capability toward cost efficiency and compute availability as more tasks become commercially viable, while maintaining that research capability and frontier model performance will remain the primary drivers of AI lab valuations.

You May Also Be Interested In

• Will U.S. data allow the Fed to remain patient?
• BoE’s Bailey warns AI models could disrupt financial markets
• Citi opens downside catalyst watch on Kroger as Walmart price war intensifies