Goldman Sachs's latest analysis highlights a major shift in China's AI agent race, positioning Tencent and MiniMax as key winners in a newly defined battlefield.
The story begins with a crucial push from the government. Chinese authorities, including the MIIT, have issued security warnings about the open-source AI agent framework, OpenClaw. These warnings highlight the risks of installing such powerful tools directly on personal or company systems, such as data leaks and unauthorized access. This has understandably made users hesitant, creating a strong demand for safer, managed alternatives.
This is where tech giants like Tencent step in, turning a security concern into a strategic advantage. First, they offer 'one-click' deployment of OpenClaw on their cloud platforms, making it incredibly easy and safe for anyone to start experimenting. Second, and more importantly, Tencent is integrating these agents directly into its super-apps like WeChat and QQ, which have over a billion users. This creates an unparalleled distribution channel, funneling massive consumer traffic directly to their cloud services.
While Tencent dominates distribution, another battle is being fought on the model front, where companies that build the core AI 'brains' compete. Here, the spotlight is on MiniMax, a recently-listed AI firm. Goldman Sachs praises its potential for rapid commercialization, partly thanks to its cost-efficient MoE (Mixture-of-Experts) architecture. This entire ecosystem is further boosted by a significant geopolitical development: the U.S. government's decision to allow the export of NVIDIA's powerful H200 GPUs to China. This eases a major hardware bottleneck, ensuring that Chinese companies have the computing power to support the growing demand for AI agents.
In essence, the competition has split into two tracks: distribution and user access (where Tencent leads) and core model technology (where MiniMax is a rising star). With Chinese tech stocks having already seen a significant price correction this year, some analysts believe the risk-reward balance is becoming more attractive, setting the stage for a potential rebound driven by this AI agent revolution.
- AI Agent: An autonomous program designed to understand a user's goals and take actions to achieve them.
- OpenClaw: A popular open-source framework that helps developers build and deploy AI agents.
- MoE (Mixture-of-Experts): An AI model architecture that improves efficiency and performance by routing tasks to specialized sub-models, much like consulting different experts for different problems.