Evaluating solo founders after generative AI

Where the rule came from

Selection criteria in venture capital and in accelerators were built in a period when a startup needed several people to exist at all. In Wasserman's (2012) dataset of high-potential startups, only 16.1% were solo-founded. A team was read as a sign of complementary skills, shared workload and commitment, and "who is your co-founder?" became one of the first questions in a pitch.

Many programmes turned the preference into a rule. Some application forms still require two or more founders.

What has changed

• More founders start alone. In Carta's data, 36.3% of new US startups in the first half of 2025 had a solo founder, up from 23.7% in 2019 (Carta, 2025). A filter that excludes solo founders now excludes a much larger part of the pipeline.
• One person can cover more. In a field experiment with 776 professionals, individuals working with AI matched the performance of two-person teams without it (Dell'Acqua et al., 2025, working paper).
• New firms are leaner. After the release of ChatGPT, new firms in areas with stronger AI skills were created with less capital, fewer shareholders and smaller founding teams (Cai et al., 2025, preprint).

What has not changed

• Key-person risk. The company still depends on one person. AI does not change that.
• Judgement and challenge. AI tools tend to agree with their user. A founder without a partner needs another source of disagreement.
• The human load. Fundraising, hiring, selling and the pressure of the role still fall on one person.
• The outcome evidence. In the first large study of launches after ChatGPT, team-based ventures still dominated the top rankings (Kim & Kang, 2026). The study measures products at launch, not companies over time, but it is a reason for caution.

Teams are not a neutral baseline either. In one survey, venture capitalists attributed 65% of the failures in their portfolio companies to problems within the management team (Gorman & Sahlman, 1989, as cited in Wasserman, 2012). The comparison is between two kinds of risk, not between risk and safety.

A note for European investors

The evidence above comes from the United States and China. I have not found public data on solo founders in European portfolios or pipelines, so a European fund or programme cannot yet check these findings against its own market.

Two things follow. First, treat the numbers as a direction. Second, your own pipeline is evidence: counting how many solo founders apply, how many are filtered out at the first screen and why takes little effort and tells you whether the question is material for you.

Questions to ask a solo founder

These are working questions drawn from the concerns in the literature. They are a discussion aid for an investment or selection committee, not validated criteria. The idea behind them is to replace "do you have a co-founder?" with "how do you cover what a co-founder used to cover?"

• Which tasks does AI do in your company today, and how do you check its work? This tests whether the capacity gap is really covered.
• Which roles are covered by paid specialists, and what do they cost each month? Advisors, contractors and fractional executives are the alternative to equity partners. They should appear in the budget.
• Who challenges your decisions? Look for a named person or group with the standing to disagree: an advisor, a board member, a peer group.
• What happens if you are unavailable for four weeks? This makes key-person risk concrete. Documentation and delegated access matter more than promises.
• Why are you building alone? A deliberate choice, with reasons, answers the signalling concern better than any credential.
• What would make you bring in a partner or a senior hire? A founder who has thought about this, and has kept equity for it, is managing the risk.
• How dependent is the product on one AI provider? A change in price or policy can reach a lean company quickly.
• What does success look like to you? A founder who wants a profitable company of ten people and an investor who needs a large exit are not aligned, however good the company is.

For accelerators, incubators and venture builders

Programmes can act on this more directly than funds. Three points are worth a review:

• Admission rules. Check whether your form or your scoring excludes solo founders by design, and whether that is still intended.
• What the cohort provides. A programme can supply part of what a co-founder gives: peers who challenge decisions, mentors with standing, and structure in difficult weeks.
• What you measure. Team size and headcount growth say less than they used to about progress.

I do not yet have evidence on which programme designs work best for solo founders. It is one of the questions of my research, and I would like to hear from programmes that are testing it.

How I work with investors and programmes

• Briefings. A 60-minute session for an investment team or selection committee on the evidence about solo founders, with time for your cases. In person in Madrid and other European cities, or by video call.
• Talks. Keynotes and panels for investor days, demo days and ecosystem events. See speaking.
• Research conversations. I talk with European investors, accelerators and incubators about how they evaluate founders. Conversations are confidential. See the research.
• Mentoring. I mentor founders in programmes at Endeavor, Unicorn Factory Lisboa, NOVA SBE and Comillas Emprende.

Write to hello@thesolothesis.com.

References

• Cai, J. J., Gu, X., Sheng, L., Xia, M., Zhao, L., & Zhu, W. (2025). AI as "co-founder": GenAI for entrepreneurship (arXiv:2512.06506). arXiv. Preprint, not yet peer reviewed. https://doi.org/10.48550/arXiv.2512.06506
• Carta. (2025). Solo founders report. https://carta.com/data/solo-founders-report
• Dell'Acqua, F., Ayoubi, C., Lifshitz-Assaf, H., Sadun, R., Mollick, E. R., Mollick, L., Han, Y., Goldman, J., Nair, H., Taub, S., & Lakhani, K. R. (2025). The cybernetic teammate: A field experiment on generative AI reshaping teamwork and expertise [Working paper].
• Kim, H., & Kang, H. (2026). Generative AI expands solo entrepreneurial entry, yet team advantage persists (extended abstract). Proceedings of the AAAI Symposium Series, 9(1), 215–216. https://doi.org/10.1609/aaaiss.v9i1.42928
• Wasserman, N. (2012). The founder's dilemmas: Anticipating and avoiding the pitfalls that can sink a startup. Princeton University Press.