The viral adoption of OpenClaw in China has a growing problem: many of the people paying for it can’t actually use it.
A WIRED investigation published March 13 interviewed half a dozen Chinese OpenClaw users and found a sharp divide between technically proficient adopters and the mass of nontechnical users who were lured in by social media hype. The latter group is spending real money — cloud server rentals, LLM API subscriptions, paid tutorials — only to discover that OpenClaw requires genuine coding skills to configure and maintain.
The George Zhang Problem
George Zhang, a cross-border ecommerce worker in Xiamen, is the profile’s central character. He rented a Tencent cloud server, bought a Kimi LLM subscription, and asked his “lobster” (the Chinese nickname for OpenClaw agents) to manage stock portfolios. It worked impressively for a few days. Then performance degraded, the agent started returning outlines instead of detailed analysis, and Zhang hit a wall: the system told him to configure API ports, a task he had no idea how to perform.
Zhang eventually gave up on stock analysis and pivoted to using OpenClaw as a WeChat content aggregator — a dramatic downgrade from the autonomous trading agent he was promised by influencer tutorials.
His experience is far from unique. Song Zhuoqun, a college student with no programming background, described the installation process alone as the most frustrating part. “There were pages full of code, and I couldn’t understand any of it,” she told WIRED. She resorted to asking ByteDance’s Doubao chatbot to generate step-by-step instructions, then copy-pasted code until errors stopped her.
The Adoption Funnel That Benefits Big Tech
The WIRED piece identifies the real economic engine behind China’s OpenClaw mania: it’s the cloud and LLM providers, not the end users, who are profiting.
Tech analyst Poe Zhao, founder of the Hello China Tech newsletter, laid out the math: “A chatbot uses only a few hundred tokens per conversation; a single active OpenClaw instance can consume tens or even hundreds of times more tokens per day.” Every new user who sets up an OpenClaw agent becomes a 24/7 paying customer for LLM API calls.
This explains why Tencent engineers were setting up tables outside company headquarters to help people install OpenClaw for free. The installation is the loss leader. The API consumption is the business model.
Alibaba, ByteDance, Minimax, and Moonshot are all running the same play — positioning their LLM APIs as the default backend for OpenClaw deployments. As we reported earlier today, Alibaba is now formalizing this into a dedicated enterprise agentic AI service.
The Binance Founder Agrees
Even Changpeng Zhao — the multibillionaire founder of Binance — weighed in on the disconnect. He posted on X that people “claim that you won’t have to do anything else after installing the lobster, but all your time after installation is spent tweaking that useless lobster that can’t do anything.”
Rain Miao, a Chinese startup founder who uses AI agents professionally, was blunter: “If you still can’t figure out how to install it after a long time, and you don’t even know how to handle the basic permissions, then you’re probably better off not installing it at all.” He pointed nontechnical users toward tools like Claude Cowork, which have received far less attention in China despite being more accessible.
Why This Matters
The WIRED report adds a critical human-cost dimension to a story that has mostly been told through the lens of viral adoption numbers and gig economy services. Thousands of nontechnical users in China are paying ongoing cloud and API costs for a tool they can barely operate — a dynamic that mirrors past technology hype cycles where the infrastructure providers captured most of the value while individual adopters absorbed the risk.
The question is whether this user frustration will cool China’s OpenClaw fever or whether the gig economy of bootcamps and installation services will absorb the gap between what users expect and what the tool actually requires.