Involution — *neijuan *(内卷) in Mandarin, literally “inward curling” — entered everyday Chinese speech around 2020. The concept was borrowed from the American anthropologist Clifford Geertz, who described a pattern in colonial Java in the 1960s: farming systems that grew more elaborate and complex without becoming more productive, absorbing ever more labor for diminishing returns.
Chinese internet users repurposed the term to describe something they recognized in their own lives — a social dynamic in which everyone works harder and harder, competes more and more intensely, yet no one actually gets ahead, because all that effort simply raises the baseline that everyone else must now meet.
In China, where decades of rapid growth have begun to slow, where graduate degrees no longer guarantee stable employment, and the housing ladder that once promised intergenerational mobility has become unclimbable for many, neijuan* *gave a name to a collective feeling that had long been present but lacked a word: the sense of running faster and faster just to stay in the same place.
In the years since Covid-19, the group I call “entrepreneurial workers” — digital laborers who are neither traditional employees nor true owners of capital — is becoming a force in the Chinese labor market. Compared to the wave of mass entrepreneurship and innovation in the mid-2010s, defined by opportunity and upward mobility, the entrepreneurial workers of this decade are caught between starting a business and working for someone else.
This new wave of micro-entrepreneurship is built on AI tools: Some people are using generative AI to write copy, create designs, run e-commerce, or produce short dramas; others are attempting, through podcasts, independent coffee shops, blogs, and small-scale creative industries, to reassemble their relationship with work and life.
If the entrepreneurial workers of the 2010s were born at the intersection of platform expansion, capital frenzy, and technological optimism, then those of 2026 are the product of a declining real estate industry, platform maturation, the spread of AI, and the normalization of risk. Both generations reflect the fact that, driven by technological change and the force of capital, labor has become more flexible, the barriers to entrepreneurship have continued to fall, and the boundary between “worker” and “entrepreneur” has grown ever more blurred. In contrast to the older entrepreneurial workers who believed that “the harder you work, the luckier you get,” the new generation of workers understands that platforms do not necessarily bring freedom, algorithms can cut off your traffic at any moment, AI may rapidly compress the premium on skills. One person is a company, using AI to write copy, create designs, edit video, handle customer service — cutting costs as much as possible, moving as fast as possible, maintaining livelihood and flexibility. They no longer dream of becoming the next Jack Ma; they hope to cover rent, social insurance, and basic living expenses, and to carve out a little room to breathe.
But perhaps because of this, their understanding of the relationship between labor and life may in fact be closer to a kind of post-growth era self-reinvention. They may not have escaped precarity, but they are beginning to ask: When success is no longer only IPOs, buying a home, and financial freedom, what else can labor bear as a life worth living?
And it is precisely in the anxiety and disillusionment of involution, that an experiment in innovation is being reincubated in China.
Unlike Silicon Valley, which is driven by financial capital, platform monopoly, and geopolitical imperatives to sustain technological leadership and capital narratives through continuous fundraising, Chinese foundation model companies including DeepSeek have been gradually feeling out a different path under the constraints of U.S. chip export controls, limited computing resources, and more cautious capital.
They cannot simply replicate the Silicon Valley model of “burn money and pile on compute” — instead, they are placing greater emphasis on finding breakthroughs through model compression, architectural optimization, engineering efficiency, and open-source ecosystems, seeking a way between lower training costs and higher deployment efficiency. This is not simply low-cost substitution; it more closely resembles a form of constrained frontier innovation forged under the twin pressures of external blockade and internal competition.
In this sense, the development of AI in China is not only the result of national industrial policy or strategic adjustments by leading technology companies, it is also deeply rooted in the broader social fabric. The entrepreneurial workers distributed across e-commerce, self-publication, short dramas, and gig platforms have for years — under conditions of severely limited resources and fierce competition — been forced to develop efficiency, rationality, and survival wisdom. They have provided the richest applications for the deployment of domestic AI models. In other words, their frugal innovations, originally regarded as individual survival strategies, are finding an unexpected resonance with the development of Chinese LLMs, and are driving an innovation ecosystem that differs from that of Silicon Valley.
Frugal innovation — a concept originating in the practical wisdom of India’s jugaad tradition of creating more value with fewer resources — may offer a reference. Today, in China, it is no longer simply a low-cost alternative from a developing country, but an institutionalized capacity capable of reshaping frontier innovation amid technology blockades, capital contraction, and high-intensity competition.
In the American mainstream mythology of entrepreneurship, the entrepreneur is typically imagined as a solitary hero, detached from family and social networks who, through talent, a spirit of risk-taking, and capital backing, changes the world through sheer personal force. Chinese entrepreneurs rarely truly go it alone. They are always embedded in the state’s policy trajectories, capital allocation, and governance logic, and they are deeply tethered to the family.
As U.S.-China technology competition intensified, and generative AI rose to prominence, the state’s role shifted from the broad mobilization of “mass entrepreneurship” to a new round of targeted organization around AI, computing infrastructure, and industrial security. Local governments began piloting AI-driven “one-person company” policy experiments — offering computing vouchers, low-rent office space, repurposed idle industrial parks, and model and data support — to absorb laid-off workers from major tech companies, generate a new form of employment, and re-embed individual entrepreneurship within the state-led AI development framework.
On the surface, this model appears to lower the barriers to entrepreneurship and encourage the “one person plus AI agents” super-individual. But it also means a vulnerability that rises and falls with the policy cycle.
Meanwhile, with the bursting of the real estate bubble, and shrinking middle-class assets, the family increasingly resembles an overextended buffer device. Many young entrepreneurs are able to set up AI startups or enter flexible employment because the risks have been transferred to their parents’ pensions, family savings, and housing assets.
As the survival strategies of entrepreneurial workers, the technical pathways of Chinese AI models, national industrial policy, and the resource-constraint wisdom accumulated over decades in the Global South converge and interact, a mode of development is gradually emerging — one that differs from Silicon Valley’s financialized expansion and also from the traditional model of imitative catch-up. It has not yet taken definitive form, and is full of contradictions. But it reminds us that the innovations of the future will not necessarily be born only where capital is most abundant. Sometimes, they may also be born where constraints are densest, competition most brutal, and the need to reinvent ways of surviving most urgent.