On March 6, nearly 1,000 people lined up outside Tencent's Shenzhen headquarters carrying laptops and hard drives, waiting for engineers to install a piece of free software. The software was OpenClaw, an open-source AI agent that hit 250,000 GitHub stars in just 90 days, a pace that took Linux years to match.
Unlike ChatGPT or Claude, which chat with you, OpenClaw actually does things. It runs locally on your computer, accesses your files, email, and apps, and autonomously completes tasks through messaging platforms like WhatsApp, Telegram, and Discord. Think of it as a personal assistant that can sort your inbox, schedule your meetings, summarize documents, and interact with other software on your behalf, all without sending your data to someone else's server.
Created by developer Peter Steinberger (who nicknamed it "the lobster" after its red logo), OpenClaw transitioned to a foundation structure after Steinberger joined OpenAI. That open governance model is part of what made what's happening in China possible.
A Whole Economy in 100 Days
The adoption in China has been less "organic growth" and more "economic event." Every major Chinese tech company is scrambling to build on top of OpenClaw. Moonshot AI launched Kimi Claw with zero-code deployment. MiniMax shipped MaxClaw as a cloud-based assistant. Zhipu AI partnered with Alibaba Cloud. Tencent released WorkBuddy, a workplace assistant that handles email triage, meeting coordination, and document analysis.
Cloud infrastructure companies are riding the wave too. UCloud Technology, QingCloud Technologies, and Hangzhou Shunwang Technology all surged over 9% on March 9, even as the broader CSI 300 Index fell 2.4%.
Then there are the individual hustlers. Feng Qingyang, a 27-year-old software engineer in Beijing, is one of many who spotted a business opportunity. The pattern is familiar from every tech gold rush: while the big companies fight over platform dominance, small operators make money by helping regular people actually set up and use the thing.
Shenzhen's Longgang district is even offering government subsidies of up to 2 million yuan (roughly $289,000) for OpenClaw-based development projects. A local government subsidizing open-source AI agent development is not something anyone had on their 2026 bingo card.
The Security Problem Nobody Wants to Talk About
Here's where the excitement outpaces the reality. Security researchers found over 40,000 exposed OpenClaw instances online, with more than 60% containing exploitable vulnerabilities. A particularly nasty flaw called "ClawJacked" allowed websites to silently hijack OpenClaw instances without user interaction, potentially stealing API keys and executing commands on people's machines.
Remember: this is software that has direct access to your files, email, and applications. A compromised instance isn't just leaking chat logs. It's a full backdoor into someone's digital life.
China's Ministry of Industry and Information Technology issued formal security alerts on March 8 and 10, warning that improper configurations posed real risks. The OpenClaw team patched over 40 vulnerabilities in February alone, but when adoption moves this fast, patches don't reach everyone quickly enough.
Speed vs. Safety, Again
Moonshot AI reported that overseas revenue surpassed domestic revenue for the first time after launching its OpenClaw product. The companies offering free computing subsidies and one-click deployment are creating real revenue through increased token consumption (every task the agent performs burns through API calls that someone pays for).
But analysts have cautioned that many of the productivity claims around OpenClaw still need verification. The gap between "impressive demo" and "reliable daily tool" remains wide for autonomous agents. When an AI agent can read your email and execute commands on your computer, the cost of a mistake is a lot higher than a bad chatbot response.
The gold rush is real. The question is whether the prospectors are building on bedrock or quicksand.