JPMorgan: China AI - Open-Weight Commercialization Playbook: Winner Takes More; Raise Zhipu to HK$2,000, Cut MiniMax to HK$300

JPMorgan argues that releasing open-weight large language models should not merely be viewed as lost monetization, but rather as a deliberate commercialization strategy shift. For competitive models, open-weight releases leverage external GPU capacity across cloud providers, inference platforms, and private deployments to drive broader adoption. Monetization can still flow through official channels, licensing, marketplace SaaS, revenue sharing, deployment support, and enterprise co-selling, particularly when customers prioritize model freshness, endpoint quality, and reliability. Conversely, for weaker models, reducing access controls accelerates price comparisons and traffic redirection away from the original provider.

This framework benefits Zhipu, given its GLM-5.2 model demonstrates strong global competitiveness. Its open-source strategy expands adoption while positioning official channels and premium variants like GLM-Turbo for quality-sensitive enterprise demand. In contrast, MiniMax faces challenges because its M3 model lacks sufficient differentiation, making it easily substitutable if users deploy the open-weight version.

JPMorgan upgraded its December 2026 target price for Zhipu (2513 HK) to HK$2,000 from HK$1,800 and maintained an Overweight rating. The firm notes that while current valuations largely reflect the company's one billion dollar annual recurring revenue guidance, the broader option value from open-weight monetization is not yet fully priced in.

Meanwhile, the target price for MiniMax (100 HK) was lowered to HK$300 from HK$400, with a Neutral rating maintained. The firm cites inadequate model differentiation and limited pricing power as primary headwinds. Ultimately, this dynamic creates a winner-takes-more environment where sustained model leadership is essential for long-term commercial success.

Key exhibits

Exhibit 1