George Zhang believed OpenClaw could make him wealthy, despite not fully understanding its viral AI agent software. After seeing a Chinese influencer showcase its autonomous stock portfolio management capabilities, Zhang, who works in cross-border ecommerce in Xiamen, decided to install OpenClaw in late February. He is among many in China captivated by the recent OpenClaw craze. Workshops teaching its use have attracted hundreds, tech firms race to integrate it, and local governments offer subsidies for related startups. Last week, images of elderly people queuing to install OpenClaw went viral online. Zhang rented a Tencent cloud server and subscribed to the Chinese large language model Kimi to interact with his OpenClaw agent, nicknamed “lobster. ” Initially impressed by its rapid, detailed market analyses, Zhang soon noticed his lobster’s performance declined to only basic outlines, and repeated attempts to get detailed reports ended with the AI endlessly claiming it was “working on it” without results. Zhang concluded OpenClaw wasn’t suited for non-programmers like him, as it required technical setup such as configuring API ports—something he couldn’t manage without step-by-step guidance. Ultimately, he stopped using it for stock trading and instead had it aggregate AI industry news to fuel a WeChat content farm. After speaking with half a dozen Chinese OpenClaw users, a clear split emerged: tech-savvy adopters see OpenClaw as productivity-transforming, while non-technical users felt misled by promises of effortless, powerful AI—only to face costly cloud server and token expenses before giving up. The real force behind China’s OpenClaw frenzy is not everyday users, but companies poised to profit. Giants like Tencent, Alibaba, ByteDance, Minimax, Moonshot, and Z. ai capitalize on AI productivity FOMO by encouraging mass adoption, benefiting from continuous LLM API usage fees. Tech analyst Poe Zhao explains that unlike chatbots consuming a few hundred tokens per interaction, an active OpenClaw can burn through tens or hundreds of times more daily, motivating Tencent to help users install it free outside their headquarters. College student Song Zhuoqun’s experience exemplifies common frustrations. Despite interning at an AI startup, her lack of coding skills made installation daunting.

Attempts to get step-by-step help from ByteDance’s Doubao chatbot were ineffective; she faced unreadable code snippets and frequent errors, leading to confusion and no real learning. Many users echoed this disappointment, highlighting a gap between OpenClaw’s hype—as accessible AI for laypeople—and its practical complexity. Binance founder Changpeng Zhao also criticized the tool, noting users spend all their time tweaking a “useless lobster” after installation rather than benefiting from it. Startup founder Rain Miao bluntly advises those struggling with installation and permissions to avoid OpenClaw, suggesting alternatives like Claude Cowork, though these attract less attention in China. Non-technical users often lack compatible or powerful computers to run AI locally, forcing them to rent cloud servers and pay for cloud-based LLMs, which increases costs. Zhang recounted spending about $30 to rent a Tencent cloud server for a year plus monthly Kimi subscriptions, with higher costs if OpenClaw performs complex, token-intensive tasks. Miao suggests experienced users can economize by delegating difficult tasks to pricier but better-performing models like ChatGPT, while offloading repetitive work to domestic AI models and running some processes locally if hardware permits. Some Chinese social media joke that unpaid interns might replace OpenClaw eventually, since OpenClaw’s reliance on costly tokens contrasts with free student labor. The key takeaway from the OpenClaw phenomenon is that ordinary Chinese users are willing to pay for AI—remarkable in a market accustomed to free software supported by data or ads. This enthusiasm drives tech firms to facilitate adoption with free installations and tutorials. Meanwhile, as OpenClaw is open-source, nearly every major Chinese tech company is developing proprietary versions—Tencent’s QClaw, ByteDance’s ArkClaw, Moonshot’s KimiClaw, and Z. ai’s AutoClaw—advertising easier installation and seamless integration with existing apps. These efforts clearly aim to lock users into their ecosystems, ensuring continued revenue from the growing AI service demand.