OpenClaw turns AI anxiety into a “compute tax” for ordinary Chinese users

AI anxiety meets consumer wallets

OpenClaw’s rise has been less about clear productivity gains and more about a modern fear of being left behind. In China’s tech discourse this behaviour is nicknamed “养龙虾” — literally “raising a lobster” — meaning consumers install and run personal AI agents largely to signal participation in the AI wave. Why pay? Because people fear missing out. It has been reported that many users burn significant sums on token consumption for routine tasks that could be cheaper or unnecessary, turning compute costs into what some call a “算力税” — a tax on anxiety.

From cloud giants to the corner installer

Big tech firms — including Alibaba (阿里巴巴) and Tencent (腾讯), alongside Microsoft and Google (谷歌) — have poured hundreds of billions into data centers and chips, funneling billions to suppliers like Nvidia (英伟达) and Broadcom (博通). It has been reported that Google’s Larry Page framed the build-out as existential, saying he’d rather “go bankrupt than lose” the race — a remark often cited to illustrate how FOMO drives capital allocation even when business models remain uncertain. At the consumer end, however, the money flows differently: cloud vendors monetize token calls, entrepreneurs earn from on-site OpenClaw installs and training courses, and many users discover only after installing that the agent mostly handles email, calendars or weather — tasks cheaper handled by other services.

Tokens, geopolitics and the uncertain payoff

It has been reported that, according to OpenRouter data, weekly calls to Chinese large language models surged back to the top globally, hitting some 4.19 trillion tokens — a spike attributed largely to OpenClaw-style usage. This surge occurs against a geopolitical backdrop: US export controls on advanced chips have made high-end semiconductors a strategic choke point, amplifying the winners in the supply chain while making compute consumption politically sensitive as well as costly. Three years into the current AI wave, it remains unclear how much real productivity ordinary users have gained, even as many have paid handsomely for the emotional reassurance of “being on the cutting edge.”

A cautionary endnote

OpenClaw may yet become genuinely transformative. Revolutions take years to show results. But history shows consumers often overestimate near-term change and undershoot long-term shifts. Not buying an OpenClaw today does not necessarily mean being left behind — just as owning a 2007 iPhone wasn’t the only ticket into the mobile internet era. What’s certain is that, for now, the fastest-growing revenue stream around AI in China is anxiety — and someone is collecting the “tax.”