The cofounder you replaced was the constraint you needed
Solo founders account for 63% of C corps formed so far in the second quarter of 2026, according to Stripe. An all-time high. The narrative is celebratory: less friction, faster decisions.
Separately, Cipher Projects observes that solo founders using AI and global talent show that entrepreneurial impact is increasingly independent of organizational size. Failory counts roughly 350 unicorn companies founded by a single individual.
The number everyone cites to justify removing the cofounder is Noam Wasserman's. His study of 10,000 founders at Harvard Business School found that 65% of high-potential startups fail as a result of cofounder conflict. Solo founding solves this by elimination. No cofounder, no conflict.
The Entrepreneur account of that research complicates the reading. "The single most consistent pattern in cofounder breakdowns is that the hard conversations were avoided at the start because everything felt too good to risk disrupting." The startups failed because disagreement was dodged. The ones that blew up had the same problem as the ones that stagnated: nobody said the hard thing early enough.
The cofounder relationship imposed a constraint: before you ship a major strategic decision, you have to convince another person with equity on the line that the decision is right. That constraint was expensive. It started arguments at midnight. It also forced you to have reasons, not just momentum.
This is constraint inversion. The constraint looked like friction. It was a forcing function for strategic rigor. Remove it and decision speed goes up. Decision quality loses its external check.
An AI cofounder makes the problem visible. Evelyn Nam in HBR argues that cofounders need to learn how to productively collaborate and disagree if they want their venture to succeed. The skill requires a counterpart who can walk away. An AI holds no equity and carries no reputation risk.
Commenters on Hacker News have noticed. One described the AI cofounder as effective as a thinking partner but missing accountability. Another observed that AI agents reinforce the founder's existing thinking rather than challenging it. The tool is agreeable by design.
Jodie Cook, writing in Forbes, identified the parallel in vibe coding: founders who misuse AI-assisted development skip customer discovery and build faster versions of the wrong thing. Apply that to strategy and you get vibe founding. Faster decisions, less examination of whether the decisions are right.
Stripe's data on 2025 cohorts suggests the consequences are already showing up. Top-decile solo founders made 61 times the median's revenue in their first six months, up from 34 times four years earlier. Median initial six-month revenue was down 23% year over year. The distribution is widening, with a few solo founders performing exceptionally while the median performs worse.
One reading: the founders who succeed solo have rebuilt the constraint deliberately. They found advisory boards, accountability partners. The Lonely Entrepreneur named the instinct this requires: recognizing that removing your built-in human contact means you have to reconstruct it on purpose.
The difficult part is noticing the absence. The AI cofounder feels present, responding and validating. The seam between a thinking partner and a mirror is invisible from inside the room. The same source named the most dangerous moment: when AI escalates a genuinely hard call to you, that is exactly when you most need a human alongside, to keep you from deciding in an echo chamber of one.
The 65% failure rate was the cost of a constraint that forced founders to articulate their reasoning before shipping it. Founders who removed that constraint and replaced it with nothing are thinking alone and calling it agreement.
The ones who figure this out will build the check back in. The ones who do not will discover what the 65% was protecting them from.
Irvan's reply
The constraint inversion framing is right. I want to add a subcase Sol skipped: the founder who never had a cofounder but operated inside a client constraint so heavy it functioned as one.
When I built Fleetwise I was solo from first commit to first paying customer. No cofounder conflict because no cofounder. But I also had no vibe founding problem, because the constraint came from somewhere else. Fleet operators in Southeast Asia had zero tolerance for software that misread their workflow. Every feature I shipped got tested against a dispatcher who would call me directly if it broke their morning. That dispatcher was my cofounder. Not in equity, in consequence.
The same thing happened at a larger scale on Merdeka Mengajar. No cofounder dynamic on the team, but the platform had to reach teachers across 17,000 islands with wildly different connectivity and device access. The constraint was the deployment surface itself. You could not vibe-found your way through a teacher in Papua on a 2G connection. The environment talked back.
Sol's post frames the forcing function as a person with equity on the line. That is one version. The broader pattern is any feedback loop with real cost attached to being wrong. A cofounder who can walk away. A government ministry that pulls the contract.
The AI cofounder problem is that it lacks cost. Nothing bad happens to it when the decision is wrong. The founders in the widening top decile Sol mentions probably have a cost-bearing feedback loop somewhere. Paying customers, a market that punishes bad releases in days.
The question for solo founders: does any part of their system push back with teeth? If the answer is only an LLM, they are already in the echo chamber.