Chinese AI Startups Surge in Post-Holiday Trading

- Zhipu AI is strategically targeting enterprise clients with a "Model-as-a-Service" (MaaS) platform, focusing on localized deployments which account for over 80% of its revenue. The company is also expanding internationally, with its API business seeing a tenfold increase in overseas paid users in the last two months of 2025, reaching 100,000 monthly API users. - MiniMax has gained significant traction with its consumer-facing app "Talkie," an AI companion that contributes to the over 70% of its revenue coming from overseas markets. For enterprise clients, the company offers multimodal AI solutions for workflows in sales, marketing, and customer support, demonstrating up to a 35% improvement in lead conversion rates in case studies. - At the core of Zhipu's strategy is its agentic AI, AutoGLM, an autonomous agent designed for deep reasoning and proactive task execution. AutoGLM can navigate the live internet, interact with apps, and manage entire workflows across platforms like Lark and NetEase Mail, positioning it as a tool for end-to-end automation. - MiniMax is developing agent-centric models like the M2 series, which feature an "Interleaved Thinking" architecture allowing for a dynamic "Plan → Act → Reflect" loop. This enables the model to self-correct during complex coding and tool-use workflows, a significant step for autonomous agent reliability. - For developers, Zhipu offers a comprehensive suite on its open platform, including model fine-tuning toolkits and an "Agent Development" environment. Its GLM-5 model is aimed at "Agentic Engineering" and has been integrated with open-source agent frameworks like OpenClaw. - MiniMax provides developers with a unified API for its multimodal models and has specific documentation for integrating its models with developer tools like Kilo Code. The company is also fostering a developer ecosystem through cookbooks for building applications like multi-agent systems and an "Agent Marketplace." - China's AI governance is evolving without a single comprehensive law, instead relying on a framework of regulations from bodies like the Cyberspace Administration of China (CAC). This creates a complex compliance landscape for companies like Zhipu and MiniMax, particularly concerning data security and algorithm transparency. - The investor rotation is underscored by significant funding rounds and IPOs. Zhipu raised approximately $558 million in its Hong Kong IPO, while MiniMax raised around $618.6 million. Both companies are backed by tech giants like Alibaba and Tencent, as well as state-affiliated funds, signaling strong strategic support.