J.P. Morgan China Internet & AI: Thoughts into Chinese New Year – A Wave of Meaningful Flagship Model Launches
Summaries are public. PDF access requires an active membership; all members have the same access. Sign in
J.P. Morgan anticipates a concentrated wave of flagship artificial intelligence model releases around the 2026 Chinese New Year, contrasting with the singular dominance of DeepSeek in 2025. Upcoming launches are expected to feature models from ByteDance, Zhipu AI, MiniMax, and DeepSeek.
The firm suggests that if DeepSeek releases an efficiency-optimized model, lower inference costs could economically enable AI integration into high-frequency consumer products rather than limiting usage to standalone chatbots. Tencent is viewed as the primary beneficiary because it has already successfully integrated third-party models into core platforms like WeChat. Vertical leaders such as Trip.com, Kuaishou, and Beike stand to benefit from a maturing open-source ecosystem that reduces adoption friction.
Conversely, the impact on Alibaba and Baidu is assessed as neutral, as efficiency gains could intensify API pricing pressure, partially offsetting their own model upgrades. Independent model providers like Zhipu AI and MiniMax are more likely to experience positive technology spillover rather than direct customer competition. Although Chinese models still trail global frontier models from OpenAI and Google, J.P. Morgan sees upside tail risks in practical workflows like coding. The firm maintains Overweight ratings on Zhipu AI and MiniMax, with target prices of HK$400 and HK$700 respectively, based on 2030 earnings expectations.