Only Chinese People Worldwide Are Frantically Snatching “Lobsters” — Doesn’t That Seem Strange?

What the “lobster” actually is

The “lobster” in question is OpenClaw — a lightweight, user‑deployed agent that gained viral fame as a seemingly omnipotent AI assistant. It began life as Clawbot, a quick hack by Austrian developer Peter Steinberger to control his laptop from WhatsApp by calling Claude; the code was open‑sourced on GitHub and quickly drew huge attention. Anthropic pushed back, reportedly arguing the project amplified and confused its Claude branding; the author then decoupled the tool from Claude, renamed it Moltbot and later OpenClaw. OpenClaw’s appeal lies not in novel cognition but in acting as a proxy: it stitches models, local hardware access, long‑term memory and reusable “skill” workflows into an automation layer.

Why the mania has been so concentrated in China

So why has the buying and course‑buying frenzy been largely domestic? Several factors converge. Culturally, Chinese tech communities have a high tolerance for rapid tool adoption and “FOMO” (fear of missing out) — people pay for courses and plug‑and‑play setups to shortcut learning curves. Economically, the rise of one‑person companies and micro teams in China makes an automation agent that can orchestrate workflows feel like an immediate productivity lever: one person can “scale” by wiring together skills. It has also been reported that Chinese AI firms and developer communities actively promoted OpenClaw derivatives, accelerating localized demand. Add regulatory and geopolitical realities — limited or costly access to some Western cloud models, tight data‑security rules and growing incentives to self‑host — and you get a market unusually receptive to a local, deployable agent.

Limits, risks and what the craze reveals

The frenzy says as much about psychology as about technology. OpenClaw is powerful as a glue layer — it can operate files, maintain persistent state and automate repetitive multi‑step tasks — but it is not a new general intelligence. Its ceiling is the model it calls; if that model falters, so does the agent. It also introduces reliability and safety risks: automated system commands, token bills and runaway loops can be costly. What the episode reveals is a pragmatic, efficiency‑hungry Chinese user base willing to pay for perceived shortcuts, and a market shaped by local platform dynamics and geopolitics rather than pure technical novelty. In short: the lobster is useful — but it is not magic.