Solo Founder Startups Surge as AI Tool Stacks Replace Teams

The one-person company moves from theory to headcount data

For years, the solo founder was seen as a compromise: someone who couldn't find a co-founder, or a hobbyist running a side project. By late 2026 that view is hard to defend. Carta's Solo Founders Report finds that solo-founded companies now make up more than one-third of new startups, more than double the 17% recorded in 20171. Another reading of Carta's data puts the share at 23.7% in 2019 and 36.3% by mid-20259. The baseline years differ between the two framings, but they point the same way, and the change is large.

The usual explanation is that AI has made building a product cheaper faster than it has made good ideas cheaper. One estimate says a founder who needed four to six employees to ship software in 2022 can now run the same operation with $300 to $500 a month in coding agents, automation tools and support bots, and that the shift took about 18 months1. The industry's biggest names have added fuel. At Anthropic's Code with Claude conference in May 2025, CEO Dario Amodei said the first billion-dollar company with one human employee would appear in 2026, and he gave that 70 to 80% odds2. Sam Altman has made a similar prediction since at least 202428.

The evidence is real, but it is uneven, and the details matter more than the slogans.

Medvi and Base44: the two cases everyone cites

Two companies carry most of the story. The first is Base44, Maor Shlomo's vibe-coding platform: users describe an app in plain language and the system builds it2. Shlomo built it alone, reached 250,000 users in about six months, and sold it to Wix for $80 million in June 20252. The company reportedly made $189,000 in profit in May 2025 alone, even after paying heavy model-token costs27. Coverage repeats the "built alone" framing almost everywhere. A recent Forbes piece, however, refers to an "eight-employee asterisk" on the story10. That detail is a reminder that "solo" often describes how a company started, not how it was staffed when it sold.

The second case is much bigger and stranger. Matthew Gallagher started Medvi, a telehealth company selling GLP-1 weight-loss drugs, from his Los Angeles home with $20,000 and more than a dozen AI tools2923. According to The New York Times, Medvi had $401 million in sales in 2025, served 250,000 customers and posted a 16.2% net profit margin, about $65 million28. It is on track for $1.8 billion in sales this year, and its only hire is Gallagher's younger brother, Elliot28. PYMNTS compared Medvi with Hims & Hers, which reported $2.4 billion in revenue with 2,442 employees and a 5.5% net margin22.

The coverage disagrees mainly on what Medvi proves. Forbes considered the proof-of-concept question settled, while calling Medvi's regulatory and competitive survival an open question29. Commentators who called it the first "billion-dollar" one-person company were corrected in public: $1.8 billion in revenue is not a $1 billion valuation25. A Taskade scorecard from March 2026 still listed the one-person $1 billion company as "not yet achieved"24.

My reading: Medvi shows something narrower and more useful than "AI replaces companies." Gallagher used AI for the front end of the business (branding, marketing, the website, customer service) and outsourced the hard parts, including doctors, pharmacies and logistics, to partners such as CareValidate and OpenLoop21. Forbes noted that the most valuable lesson may be for those infrastructure providers, which have now shown they can support a $1.8 billion business29. The one-person company here is mostly a coordinator sitting on top of other companies' payrolls.

What solo founders actually earn

Outside the outliers, the revenue picture is smaller and steadier. The same names appear again and again. Pieter Levels runs NomadList, PhotoAI, RemoteOK and InteriorAI with no employees, and his public revenue dashboard shows more than $3 million a year1. Danny Postma's HeadshotPro is reported at roughly $300,000 a month, run from Bali2, although one tracker lists the team as having grown from solo to three people24.

The more telling examples are lower down. On Indie Hackers, Jason Zigelbaum says he runs Zigpoll with no co-founder, funding or sales team, and that it reached about $125,000 in monthly recurring revenue (MRR) by mid-2026, after two years of little traction15. Jack Friks built Post Bridge, a social-media scheduling tool, to about $35,000 MRR, doing nearly all the work himself apart from a few contractors19. Jason McCreary's Laravel upgrade service, Shift, earns more than $50,000 a month, though he says growth has slowed as developers adopt AI20. That last point is a useful warning: the technology that lets solo founders build can also erode demand for what they sell.

The overall distribution is the part promotional coverage tends to leave out. One overview says the median active indie hacker makes under $1,000 a month, with only the top 10% above $10,000 MRR14. Taskade, which sells tools to this market, estimates that most solo founders earn $3,000 to $5,000 a month and that about 2 to 3% reach $1 million in annual recurring revenue (ARR)24. Figures from a "Solo Founder Index" covering 2,500 companies look stronger at first: 28% of AI-using solo founders reportedly reach $100,000 ARR within a year, against 11% without AI, and 4.2% reach $1 million ARR within two years, against 0.8%9. Even if those numbers hold up, the post that reports them concedes the absolute numbers are small9. AI seems to improve the odds of success, but the odds are still low.

The indie hacker tech stack has settled down

The tooling has converged more than the hype suggests. A typical 2026 indie stack is Next.js with Tailwind and Shadcn on the front end, Supabase or PlanetScale for the database, Vercel for hosting, Clerk or Auth.js for login, and Stripe or Paddle for billing14. One detailed guide adds Cursor or Claude Code for coding, PostHog for analytics, Sentry for errors, Resend for email and Crisp for support, and puts the total at $30 to $550 a month13. Cursor Pro and Claude Pro each cost about $20 a month1. Lovable, Bolt.new and Replit Agent have let people with no engineering background launch software, and some of those products are reported to reach $5,000 to $50,000 MRR1.

Two details complicate the "$20 a month replaces a developer" pitch. Levels, the best-known solo operator, deliberately uses plain PHP on a single cheap server1, and Zigpoll's founder describes a stable Express, Mongo and React setup15. The founders with the longest track records tend to choose simple, proven tools over the newest ones. Costs can also rise unexpectedly: Friks moved storage to Cloudflare R2 after his Supabase data-transfer bills reached about $1,000 a month19.

Adoption is also shallower than the toolkits imply. A 2026 Clarify Capital survey found that 63% of solo founders use AI, but only 8% use it across everything from strategy to daily operations10. ChatGPT led with 42% adoption, followed by Gemini at 27%10.

The costs: security gaps, broken chatbots and isolation

The one-person model has costs that its strongest promoters tend to skip. Kosovo-based developer Lorik Morina built a business around them. His scanner, SecureVibing, looks for exposed API keys and misconfigured Supabase row-level security, mistakes he says are common in vibe-coded apps12. In one example he gives, a free user could change their own database record to "Premium"12. When code is cheap to write, mistakes are cheap to ship too.

Medvi shows the operational side of the same problem. Its chatbot sometimes invented drug prices or suggested products the company didn't sell. When customers asked for a person, the calls went to Gallagher's own phone, which led to more than 1,000 support calls21. Over time he added outside partners for operations, legal work and accounting, and he told the Times he wants to hire people partly because he is lonely21. Other analysis points out that having only two people concentrates a great deal of responsibility and operational risk26.

The model also has structural limits. Forbes argues that it struggles with enterprise sales, regulated industries and AI costs that grow with usage, and that while angel investors increasingly back solo founders, venture capital firms remain cautious1. A Series A round is still structurally harder to raise without a team1. Y Combinator, though, has listed tools that give solo founders "100x" leverage as a priority category, and Sequoia has reportedly begun factoring AI-agent leverage into how it evaluates startups1.

Bottom line

Much of the excitement comes from companies selling the stack: accelerators repositioned as "AI-native company builders"7 and workspace tools promising agent-run org charts24. That doesn't make the trend false. Carta's numbers and the revenue disclosures on Indie Hackers stand on their own.

My conclusion: AI has made the profitable solo software business, earning somewhere between $10,000 and a few hundred thousand dollars a month, much more achievable, and that is the real change. The billion-dollar solo company is less convincing as a pattern. Its best example so far relies on outsourced infrastructure and a fast-growing drug category, and its founder is already adding people. The most likely result is many more small, profitable companies, not a wave of one-person giants.