Goldman Sachs Names Top Chinese AI Models, Backs Zhipu

Goldman Sachs Names Top Chinese AI Models, Backs Zhipu

Goldman Sachs has identified three Chinese artificial intelligence models as its preferred picks in the increasingly competitive domestic AI landscape, with only one currently available to public investors.

The investment bank initiated coverage on Hong Kong-listed Zhipu—also called Knowledge Atlas—with a price target of HK$1,880 ($239.83), representing nearly 15% upside from its closing price. Zhipu has surged since its January listing in Hong Kong, gaining additional momentum after its open-source GLM-5.2 model demonstrated performance comparable to Anthropic's Fable 5 across multiple benchmarks, according to CNBC.

Goldman assigned Zhipu a neutral rating despite the upside target. The firm's two other preferred Chinese AI model companies—DeepSeek and ByteDance—remain privately held.

Performance and Market Position

Goldman's analysis evaluated models across several dimensions including time to market, arena scores, valuation, and pricing. The research also assessed AI video generation capabilities, where ByteDance ranked highest.

Zhipu's GLM and DeepSeek's models generally outperformed offerings from Alibaba, Tencent, and Minimax, particularly in time to market and arena scores. The analysts noted that Zhipu's extensive adoption among coders should enable sustained model improvements and reinforce its leadership position in enterprise and coding applications within China.

Market performance reflects this divergence. Over the past 60 trading days in Hong Kong, Zhipu shares climbed 70% while Minimax plunged more than 70%. Alibaba declined nearly 10% and Tencent dropped approximately 5% during the same period.

Why it matters

Chinese AI models are reaching a critical inflection point where open-source and open-weight approaches deliver intelligence performance approaching global proprietary models at significantly lower costs. This shift matters for enterprise buyers globally, particularly as agentic AI applications drive explosive demand for cost-effective models. The competitive landscape also highlights how access to computing resources—shaped by U.S.-China regulations, balance sheet strength, and inference efficiency—will determine which players can sustain leadership as the technology advances.

Computing Access as Swing Factor

Goldman analysts emphasized that Chinese AI open-source and open-weight models are reaching "a critical point of intelligence performance vs. global proprietary models." They identified agentic AI as a driver of explosive demand for value-oriented models at the lower end of the market.

The analysts highlighted that access to computing resources will be a determining factor in competitive outcomes, with U.S.-China regulations, balance sheet capacity, and inference efficiency playing key roles.

Details were first reported by CNBC.