Tiny Teams, Big Startups: The GenAI Co-Founder Turning Lean Ventures into a New Entrepreneurial Boom
Introduction: why this matters to any aspiring founder
What if the barrier to starting a business could shrink overnight because you had a “digital co-founder” by your side? Enter GenAI—the family of generative AI tools that can draft a business plan, write code, spin up marketing content, crunch data, and even handle customer interactions. A lot of people have talked about AI boosting productivity for existing firms, but what about starting new ones from scratch? A recent study looking at China’s startup surge after the November 2022 ChatGPT release suggests GenAI isn’t just making existing businesses more efficient—it’s changing the very economics of entry. In short: GenAI can help individuals with few resources start leaner, smaller startups, and it does so most strongly in places where people already know AI-related stuff.
In this blog, we’ll break down what the researchers did, what they found, and what it means for would-be founders, investors, and policy folks. No heavy math, just the story behind the numbers and the practical takeaways you can use.
What the study was trying to answer
- Does GenAI diffusion spur new firm creation? And who benefits most?
- Through which channels does GenAI affect entry costs, access to capital, and the need for labor?
- Are the effects uniform, or do they depend on firm size, industry, and local AI know-how?
- How robust are the findings to different checks and alternative explanations?
The big idea: GenAI as a “co-founder”
GenAI’s magic isn’t that it builds a product by itself, but that it lowers the frictions involved in starting a firm. Think of GenAI as a partner who can code, draft, research, market, and plan—reducing the need for a big, expensive team at the outset. The researchers call this a shift toward “lean ventures” and a redefinition of the minimum viable scale for new businesses.
Data: a nationwide, high-resolution view
- Location data: They map every new Chinese firm to a fine-grained hex grid (about 5 square kilometers per cell) and track registrations from 2021 through 2024.
- Size split: Firms are classified as “small” if registered capital is under 1 million RMB; the rest are “large.”
- AI capital proxy: Before ChatGPT, the researchers used AI invention patents filed between 2010 and 2019 as a proxy for AI-specific human capital in a grid. That pre-existing AI know-how helps identify where GenAI diffusion would land most effectively.
- What they compare: They look at grids within the same city, before and after the ChatGPT shock (November 2022), comparing grids with high pre-existing AI patent activity to those with little or none.
Key findings, in plain terms
- A national surge in new firms after ChatGPT, concentrated in AI-rich grids
- Grids with more AI know-how before 2020 saw a noticeable uptick in new firm formation after ChatGPT’s release.
- On average, high-AI grids gained about five extra new firms per grid per quarter compared with their lower-AI neighbors.
- When you sum across all high-AI grids and eight quarters after the shock, that adds up to roughly 51,000 extra firms per quarter—roughly 6% of all firms created nationwide in that period.
- The surge is all about smaller, leaner startups
- The bump in entry comes entirely from small firms; large-firm entry actually declines in high-AI areas after ChatGPT.
- This pattern is consistent with GenAI slashing fixed costs and enabling startups to get off the ground with smaller teams and less external funding.
- Founding teams, capital, and structure get smaller
- Serial entrepreneurs (people who previously started firms) were less likely to dominate new entries after ChatGPT; first-time founders did most of the entering.
- New firms in AI-rich grids tend to have fewer shareholders and smaller founding teams.
- Even among serial entrepreneurs, new ventures tend to be smaller than their past ventures post-ChatGPT.
- Sectors and AI-relevance matter
- The strongest entry gains appear in AI-downstream sectors—areas where GenAI can be readily embedded into products and services (retail, business services, online platforms, digital services).
- Traditional, capital-intensive sectors like construction and manufacturing showed weaker or even negative effects.
- They dug deeper into industry heterogeneity by looking at AI-upstream vs AI-downstream activity and three AI-relevance dimensions (upstream AI development, downstream AI applications, and entrepreneurship-helpfulness for firms).
- Downtowns and online/creative sectors benefited more; upstream hardware and infrastructure sectors didn’t see the same boost to new entries.
- Mechanisms at play: what’s really happening under the hood
- Experience and labor: GenAI substitutes for some managerial and specialized labor, making it easier for first-time founders to launch.
- Financing friction: Firms could start with fewer shareholders, alleviating the need to gather a lot of capital upfront.
- Labor organization: Founding teams could be smaller, as GenAI supports many early-stage tasks that used to require a larger team.
- Robustness and credibility: the science behind the claim
- The authors ran multiple placebo tests: replacing AI patents with non-AI patents (to test for general innovation effects), residualizing entrepreneurship (to account for pre-existing differences), randomizing the “high-AI” label (to check for spurious spatial correlations), and matching AI-active grids with geographically close AI-inactive grids.
- In most placebo checks, the AI-specific effect vanished or weakened substantially, supporting the idea that AI-specific human capital is the key driver.
- Excluding the top-tier provinces (Beijing, Shanghai, Guangdong) still left the core result intact: lean, small-firm entry rose in AI-exposed grids, with large-firm entry falling.
- They also tested varying the cutoff for what counts as a “small” firm and found the results were robust.
Why this matters: implications for founders, firms, and policymakers
- Democratizing entrepreneurship
- GenAI appears to democratize who can start a company. If you can cobble together a lean team of one or a few people with GenAI tools, you can enter markets you previously would have needed a larger operation to tackle.
- A shift in the startup ecosystem
- Expect more “AI-enabled” startups in digital services, marketing, software, and other knowledge-work domains.
- The jobs and skills that matter at the outset may shift away from “hiring a lot of specialists” to leveraging AI tools that augment a small founding team.
- Regional and policy considerations
- Areas with pre-existing AI human capital gain the most, underscoring the importance of local talent pipelines, AI literacy, and access to AI tools.
- For policymakers, this suggests targeted support for AI-adjacent entrepreneurship—e.g., cheap compute, accessible AI tooling, and training—could yield outsized effects in terms of startup formation and local innovation.
A closer look at how GenAI reshapes firm formation
1) The “co-founder” idea in practice
- GenAI acts like a silent partner that can draft plans, prototype products, write code, draft marketing materials, and handle repetitive tasks.
- With such a partner, a solo founder or a tiny team can move from idea to minimum viable product faster than before, reducing the required size of the founding team.
2) Who benefits the most?
- Small firms and first-time founders see the biggest boost.
- Large firms, which typically rely on more extensive teams and established structures, see smaller gains or even declines in entry in AI-exposed zones.
3) What kinds of tasks loosen up?
- Tech-enabled and service-oriented tasks: product prototyping, content creation, coding, data analysis, marketing, and customer interaction.
- Replacing or augmenting routines that used to require several specialized hires.
4) Where the action is most intense
- AI-downstream industries: sectors that can readily incorporate GenAI outputs into their products and services.
- Retail, business services, and digital platforms show strong responses.
- Upstream AI infrastructure (hardware, data centers, core model development) remains more capital-intensive and less prone to immediate entry.
Key takeaways and practical ideas for readers
- GenAI can be a powerful equalizer for early-stage entrepreneurship.
- If you’re launching solo or with a tiny team, GenAI tools can substitute for a broader staff, lowering the fixed costs of starting a company.
- The benefits are not uniform: regions with more AI know-how and access to AI talent see bigger effects, and downstream, creative, service-oriented firms gain more than hardware or heavy-manufacturing sectors.
- Expect a shift in startup patterns: more lean, smaller ventures, fewer co-founders per startup, and possibly less serial entrepreneurship in AI-rich zones as GenAI lowers the bar to entry.
- For policymakers and city planners, investing in AI literacy, access to tools, and local AI talent pipelines could spark more new ventures and diversify the startup ecosystem.
If you’re thinking about starting your own venture, here are some practical prompts and strategies inspired by the study
- Leverage GenAI for the core early tasks
- Use GenAI to draft your business plan, create a lean product roadmap, and prototype marketing copy and landing pages.
- Consider a one- to two-person founding team: one technical founder plus a founder focused on product-market fit and operations, both supported by GenAI.
- Focus on AI-downstream opportunities
- Look for problems where a GenAI-enabled workflow can dramatically cut costs or time-to-market—areas like digital marketing services, content creation, software-enabled services, or data analytics for small businesses.
- Build with the minimum viable scale in mind
- Use AI to test multiple small experiments quickly, measure outcomes, and pivot—rather than trying to build a full product before market feedback.
- Invest in local AI know-how
- If you’re in a region with strong AI talent, you’ll have an advantage in recognizing, adopting, and deploying GenAI tools for early-stage ventures.
- Keep an eye on founders’ experience
- If you’re a first-time founder, GenAI tools may help compensate for limited prior startup experience, but be mindful of the learning curve and partner with mentors or peers who can complement your skill set.
Limitations and a note on the bigger picture
- The study is China-focused, using a rich administrative dataset and a sharp exogenous shock (ChatGPT release), which makes the setting especially informative. Still, dynamics could differ in other countries with different regulatory environments, startup cultures, and AI ecosystems.
- The measure of AI human capital relies on patents from 2010–2019 as a proxy. It’s a reasonable indicator of local AI knowledge, but it isn’t a perfect capture of all the capabilities that might influence GenAI adoption.
- The long-run effects—how this reshapes productivity, job quality, and industry structure beyond the initial entry surge—still need more time to unfold and study.
Conclusion: a new entrepreneurial landscape in the making
The key message from this research isn’t just about more startups after a breakthrough technology. It’s about a shift in what it takes to start up—and what counts as a viable startup in the first place. GenAI is nudging the economy toward leaner, more nimble ventures, especially in places where AI know-how is already present. It’s not replacing human ingenuity; it’s expanding who can participate and how early-stage teams can operate. If you’ve ever thought that starting a company required a big budget, a long hiring queue, and a large risk appetite, GenAI might be changing that calculus for you.
Key Takeaways
- GenAI diffusion after the ChatGPT shock led to a meaningful uptick in new firm formation in China, concentrated in grids with pre-existing AI-specific human capital.
- The surge is driven entirely by small firms; large-firm entry declines in AI-exposed areas, signaling a shift toward leaner ventures.
- Founding teams got smaller, with fewer shareholders and less managerial labor at inception, suggesting GenAI lowers fixed costs and substitutes for early-stage talent.
- The strongest effects occurred in AI-downstream, knowledge-based sectors where GenAI can be easily embedded into products and services.
- The entrepreneurship boost predominantly benefited first-time founders, not serial entrepreneurs, indicating GenAI helps newcomers overcome entry barriers.
- Robustness checks, placebo tests, and matching exercises support the conclusion that AI-specific human capital—not broad innovation activity—drives the post-ChapGPT surge in firm formation.
- The study highlights the importance of regional AI talent and access to AI tools for fostering startup dynamism, with implications for entrepreneurs, investors, and policymakers seeking to stimulate job creation and innovation.
If you’re curious about prompting GenAI for your own venture or just want to talk through how a “digital co-founder” could fit into your startup plan, I’m happy to brainstorm ideas or help tailor a prompt strategy that aligns with your industry and resources.
