Morgan Stanley sees 3x growth for China AI, lifts Zhipu target to HKD990

Morgan Stanley has significantly increased its valuation for China's top artificial intelligence firms, forecasting that the sector's annual recurring revenue (ARR) will reach between $1 billion and $1.5 billion by the end of 2026. The bank's report cites rapidly improving model intelligence and inference costs that are just 15% to 20% of US counterparts as key drivers for 3x to 5x annual growth.

"Foundation models continue to scale, benefiting from US computing power and China’s engineering efficiency," the Morgan Stanley research report said. The analysis points to a surge in token usage and pricing power as signals of non-linear revenue growth for high-intelligence-density models from companies like Zhipu, MiniMax, and Alibaba.

In response to the forecast, the bank raised its price target for Zhipu (02513.HK) by 77% to HKD990 from HKD560 and lifted its target for MiniMax (00100.HK) to HKD1,100 from HKD990, maintaining Overweight ratings on both. The report projects the market could further expand to between $2.5 billion and $5 billion by the end of 2027.

The bullish outlook rests on the idea that Chinese AI labs are achieving near-parity with US peers in model performance while maintaining a structural cost advantage. This allows them to compete aggressively on price and capture a significant share of the growing demand for AI applications, creating a new growth vector for the country's tech sector.

DeepSeek V4 Highlights Engineering Edge

The recent launch of DeepSeek's V4 model exemplifies the engineering efficiency highlighted in Morgan Stanley's report. According to DeepSeek, its V4 Pro model outperforms all open-source competitors in reasoning tasks and was notably adapted to run on Huawei's Ascend AI chips, a technical shift away from a sole reliance on Nvidia processors. The model's ability to support a one-million-token context window at a low cost is particularly relevant for the complex, long-context agent tasks that are expected to drive enterprise adoption.

While DeepSeek's performance claims await broad independent verification, the model quickly became the top trending model on the Hugging Face platform after its release. The move to train and deploy on domestic hardware from Huawei also shows progress in building a resilient, local supply chain, a crucial factor as US export controls continue to shape the competitive landscape.

China's Tech Efficiency Playbook

The strategy in AI mirrors a pattern seen in other advanced technology sectors, such as humanoid robotics, where Chinese firms have leveraged supply chain integration to drive down costs and accelerate production. Companies like Unitree and Agibot are projected to ship tens of thousands of humanoid robots in 2026, with manufacturing costs dropping 40% per year, far faster than analysts expected. This is largely due to the same vertically integrated supply chains for components like joint actuators that serve the country's dominant electric vehicle industry.

While the US leads in venture funding and ambitious projects from firms like Tesla and Figure, China is winning on volume and cost. This playbook—combining rapid iteration, supply chain control, and aggressive pricing to scale production—is now being applied to frontier AI models, creating a distinct competitive advantage.

For investors, Morgan Stanley identifies Zhipu and MiniMax as key picks, with Alibaba (09988.HK) also favored for its comprehensive AI capabilities. The bank's bull case valuations see Zhipu and MiniMax rising to HKD1,730 and HKD2,480 per share, respectively, suggesting significant upside if the efficiency and growth narrative plays out as projected.

This article is for informational purposes only and does not constitute investment advice.