Raising a Lobster: China's Agentic AI Gamble

13 April 2026  |  

Sruthi Kalyani
Geopolitics
Economy

Image
 

Somewhere in China, someone right now is 'raising a lobster'. It could be a veteran in Shenzhen experimenting with newly installed software, or an office worker in Beijing relying on AI to draft emails, while quietly worrying that without mastering AI, she might soon be replaced. The centre of all of it is OpenClaw, an open-source AI agentic tool – a software that gets things done and completes complex tasks with minimal or no human intervention. Unlike a chatbot that answers questions and waits, OpenClaw acts independently. Nicknamed the 'little lobster' (xiaolongxia), it has, within weeks of its release, become the most talked-about tool in the world's most consequential AI market. China neither created nor controls it, and its government has urged caution about the risks. None of that has slowed the obsession – China has already surpassed the United States in total adoption. And it raises critical questions about what this runaway adoption tells us about what a government will control, and what a society will demand of people who cannot afford to be left behind.

Why China adopts faster

China's embrace of agentic AI is not really to go along with the global trend. It emerges from three structural pressures. The first is fundamentally demographic. Youth unemployment in China has remained elevated, and the Communist Party of China’s (CPC) unspoken social contract of economic mobility in exchange for political compliance is under measurable strain. Agentic AI, in the Party's framing, comes as a productivity multiplier capable of decoupling economic output from an ageing workforce.

The second is corporate desperation. OpenClaw seemingly offers a path to AI relevance that does not depend on model supremacy. For cloud giants like Tencent who have built vast compute infrastructure but cannot generate sufficient demand from conventional chatbots, agentic AI is the catalyst that finally makes the capital expenditure profitable. By embedding an OpenClaw agent directly into its messaging and social media app – WeChat – Tencent has shifted the competitive battlefield from whose AI is capable to whose AI is more ubiquitous, at least within China.  Notably, just on the day it announced three OpenClaw products simultaneously, Tencent's stock rose nearly seven percent. Other Chinese companies including ByteDance, Alibaba and Baidu have launched domesticated versions of OpenClaw, thereby replacing the foreign model with state-approved Chinese large language models, keeping all data within the Party’s jurisdiction.

The third pressure is the need to maintain a permissive regulatory climate. On a national level, the Chinese state has formalised this logic through its ‘AI Plus’ initiative, which mandates 70 per cent AI penetration across key industries by 2027 and 90 per cent by 2030. China's 15th Five-Year Plan elevates embodied intelligence as a key future industry, treating large‑scale deployment of AI agents in physical systems as a key vector of the state’s industrial upgradation. In March 2026, Premier Li Qiang's Government Work Report, for the first time, referenced ‘intelligent agents’ in its official text – a signal that agentic AI is a state priority. Local governments have, since then, actively offered grants and subsidy schemes promoting agentic AI adoption. China's permissive environment makes rapid deployment frictionless in ways that Europe and the United States simply cannot match. While, the European Union’s AI Act provides a unified risk-based framework that  covers autonomous agent deployments, China has no comparable overarching legislation yet.

The Party-State's Dilemma

And yet the latest craze has collided almost immediately with the CPC's most fundamental instinct – control. Attackers have already used OpenClaw's public marketplace to distribute malware disguised as legitimate tools. While CNCERT, China's top cybersecurity body, has issued security guidance for users, cloud providers and developers, state-sponsored labs are running vulnerability audits and building detection tools for the agent. But none of this resolves Beijing's deeper concern – OpenClaw raises serious questions about a potential breach of China's stringent data localisation laws, possibly pushing their data to foreign servers that may be beyond the Party's reach.

The deeper geopolitical contest is now in plain view. While American firms pour hundreds of billions of dollars into giant data centres chasing Artificial General Intelligence, China is focused on adoption and physical integration that embeds agentic AI into daily life. China’s Premier Li Qiang has positioned China as the AI partner of choice for the Global South, offering open-source models and shared infrastructure as an alternative to American proprietary systems. OpenClaw, domesticated and redistributed through Chinese cloud providers, fits neatly into that offer. Countries that cannot access Western models may access Chinese-hosted agents instead.

This lobster craze, however, shall pass. What shall not pass is the transformation it has accelerated. Agentic AI has moved from experiment to infrastructure faster than any governance framework was prepared for.

India's own adoption has been quieter than China's but is accelerating, particularly in the IT sector. The Digital Personal Data Protection (DPDP) Act, passed in 2023, was not built for a world where an autonomous  agent accesses personal data without the user's knowledge. The question is whether India can build adequate guardrails before adoption outruns oversight, as it already has in China. China's bet is that it can harness the productivity revolution of autonomous AI while maintaining the political control on which its legitimacy depends. This gamble is a contradiction in terms – the very openness that makes agentic AI powerful is what the  Party-state most fears.

This article was originally published in The Tribune as part of an arrangement with the Centre of Excellence for Himalayan Studies


About the Author: Sruthi Kalyani is a Policy Fellow at the Centre for Society and Policy, Indian Institute of Science, Bengaluru