NBD Hot Comment | "Farming Lobsters" Is Not "Awarding Lobsters" — True Innovation Doesn't Need "Force-Feeding"

Daily Economic News Reporter: Du Hengfeng    Editor: Yang Jun

Since Shenzhen Longgang District announced the OpenClaw “Lobster Ten Rules,” Wuxi High-tech Zone, Changshu City in Suzhou, Hefei High-tech Zone, Hangzhou Xiaoshan District, Nanjing Qixia High-tech Zone, and Jiangning District have followed suit. The “lobster farming” craze has rapidly spread from the geek community and capital markets to local government investment promotion efforts. While people marvel at local governments’ deep understanding of new technologies and precise grasp of new industry development opportunities, some common issues in investment promotion deserve close attention.

“Lobster farming” should not become “Lobster Rewarding.” The most direct, noticeable, and “real” subsidies are often highlighted in headlines. Some local governments have boldly listed the maximum subsidy amounts in their press releases. The earliest maximum reward in Shenzhen Longgang District was 4 million yuan, followed by versions offering 5 million, 6 million, 10 million, and even 20 million yuan.

Regarding office space incentives, Shenzhen Longgang District offers OPC (one-person company) up to 18 months of free office space. Subsequent regions have increased support, offering “up to 2 years of free dedicated workspace, with water, electricity, property, and internet fees waived,” or “up to 3 years of rent-free office space support,” and even “up to 5 years of rent subsidies, with a maximum of 3,000 square meters annually,” among others.

“Lobster farming” and OPC support policies are closely linked. Shenzhen Longgang District provides “up to 100,000 yuan household registration subsidy + up to 2 months of free accommodation,” while other regions offer “up to 120,000 yuan living subsidy,” “30 days of free office, accommodation, and dining + high-speed rail subsidy,” “up to 36,000 yuan annual housing rental subsidy + 300,000 yuan employment subsidy + up to 6 months free stay at talent stations,” and even “up to 2 million yuan home purchase subsidy.”

Shenzhen Longgang District proposes up to 10 million yuan in equity investment support. Subsequent policies uniformly include “equity investment” as standard, with some offering OPC-specific credit products and 50% loan interest subsidies.

These reward and support policies are somewhat reasonable. New business opportunities in their nascent stage face high costs and risks; appropriate government support can lower entrepreneurial barriers and risks, allowing innovation to flourish. However, these policies should focus on “ensuring basic needs” and not turn into competitions of scale or coverage. Fiscal resources are scarce and precious; more impact can be achieved by investing in employment, education, and other public welfare areas.

“Lobster farming” should not become “Lobster Picking.” Many see OpenClaw as a “DeepSeek moment” for intelligent agents, eager not to miss out even if it means making mistakes. OpenClaw indeed demonstrates AI’s potential to move from guiding humans on what to do to independently performing tasks, but the development of intelligent agents is still in a very early stage. Even with OpenClaw, issues like deployment difficulty, compatibility, and high token consumption exist, not to mention serious security risks.

For example, competitors of ChatGPT can quickly catch up or even surpass in certain areas. The technical barrier for AI applications like OpenClaw is much lower, and in the future, better intelligent agent tools may emerge. Policy support for new technologies should encourage foundational innovation rather than selecting winners prematurely.

Clearly, OpenClaw is far from being the ultimate solution for intelligent agents. Favoring OpenClaw in support policies is not “lobster farming” but “lobster picking,” which could crowd out other innovative AI developments. During the early stages of technological emergence, this is highly detrimental to innovation.

I also note that some regions include other intelligent agents in their support policies, but these mentions are brief. From an entrepreneur’s perspective, they are more likely to choose OpenClaw to receive rewards than less-known intelligent agents. The answer is obvious.

“Lobster farming” should not become “Fattening Shrimp.” The success of a new technology is always driven by market choice. Market selection considers availability, cost, safety, and other factors. The development of intelligent agents also follows these principles. Comprehensive subsidies for computing power, data, models, office space, and living expenses do lower entrepreneurial barriers, but products built on such support often have artificially suppressed costs. Without subsidies, their commercialization stories become much harder. If many “fattening shrimp” appear simultaneously, the market will ultimately weed them out through brutal淘汰, incurring huge economic costs.

The case of the steam engine best illustrates how new technology succeeds. Early steam engines were heavy and inefficient, consuming大量煤炭. For coal mine owners, coal was cheap, so high energy consumption was acceptable, but for textile factories, such steam engines were uneconomical. It wasn’t until Watt improved the steam engine, significantly reducing energy consumption, that it could be widely applied in textiles, mining, transportation, and other industries.

AI development must also address cost issues, primarily related to computing power, which in turn depends on energy. Energy has become a critical bottleneck for AI development. Intelligent agents consume enormous amounts of computing power; solving cost issues is essential for widespread application. However, subsidies reduce entrepreneurs’ focus on this problem, much like coal mine owners with endless coal supplies who have no incentive to improve steam engines.

Daily Economic News