Fired for Using It, Fired for Not: China's AI Mandate Economy Is Now Measurable
Peter Steinberger, OpenClaw’s creator, described the situation at a TED-format event this week: “In China, installing OpenClaw is called raising lobsters. Thousands of people were lining up at the Tencent office in Shenzhen to get their lobster installed. Shenzhen even gives out subsidies for people running businesses on OpenClaw.”
Then the harder fact: “I met an entrepreneur in China who showed me a spreadsheet. Every employee, every day, one task automated by OpenClaw. If you miss too many days, you’re fired.”
Fortune and the Los Angeles Times confirmed the broader picture, describing crowds in Beijing and Shenzhen gathering at tech company offices for help setting up AI agents on their laptops. The scene repeated for weeks.
Steinberger’s summary: “Fired for using it, fired for not using it.”
Two Adoption Regimes, One Technology
In the United States, the deployment conversation is happening mostly at the executive and policy layer. Companies are announcing AI spending, cutting headcount, and publishing productivity metrics. Individual workers are encountering AI tools at uneven rates depending on their employer and role.
In China, the pressure is being systematised at the employee level. Daily task automation quotas. Spreadsheet tracking. Municipal subsidies for businesses that adopt AI agents. The government’s role is explicit: Shenzhen treating OpenClaw adoption as an economic development priority is not a market signal, it is a state signal.
The Ipsos polling firm found China is the most AI-enthusiastic country in the world by self-reported survey. The enthusiasm is partly genuine and partly structural — the incentive architecture is different.
The Qwen Layer
Alibaba’s strategy compounds the adoption dynamic. Qwen is not positioned as a standalone model. It is being embedded across Alibaba’s consumer and business ecosystem: shopping, payments, maps, travel, office tools, education, healthcare. Medical researchers in China are using it as a workflow layer — gathering papers, sorting evidence, framing mechanisms, drafting research-style explanations — inside tools they already use for other tasks.
The effect is that AI contact happens without a separate adoption decision. Users in Alibaba’s ecosystem encounter Qwen capability inside applications they already depend on, not as a distinct product requiring a deliberate install. The AI becomes ambient.
That is a different bet from OpenAI’s model: OpenAI is building a research assistant that users choose to consult. Alibaba is building a surface that is already where the work is. Whether that distinction matters in the long run depends on which usage pattern proves more durable — but in 2026, the surface bet is producing scale.
The Divergence Will Show Up in Productivity Data
The gap between Chinese and American AI adoption rates is currently documented anecdotally and through self-reported surveys. By 2027, it will show up in enterprise productivity data. Companies that have been running daily automation quotas for eighteen months will have measurably different output structures than companies where AI adoption has been voluntary and uneven.
The UCF humanities graduates who booed an AI mention at commencement last week are entering the labour market. The Chinese tech company employees who miss their daily automation quota face termination. These are not equivalent positions, and the difference is not primarily about which country has better AI models.
The race Steinberger described is not a race to the best model. It is a race about which society is willing to restructure itself around AI the fastest — and which bears that cost on whom.