Zero Solo Unicorns Yet, But $1M Solo Companies Have Doubled

Zero solo unicorns so far, but $1M-revenue one-person companies have doubled

In May 2025, Anthropic CEO Dario Amodei made a prediction: “a billion-dollar company with a single employee will emerge in 2026.” He put the odds at 70–80%. We’re already halfway through that year now, so how’s the scorecard looking? Solo unicorns: still zero.

But The Wall Street Journal recently ran an intriguing piece. Something Amodei didn’t predict showed up in the payment data instead. The number of “zero-employee companies” clearing $1 million in annual revenue has doubled in two years. Those crossing $10 million have nearly tripled. There’s no billion-dollar one-person company yet — but a one-person company pulling in $1 million has become a lot more common.

How much should we trust these numbers? What’s actually changed in this cycle isn’t the success of solo companies itself — it’s that we now have material other people can use to verify those success stories. I’ve sorted this material into four tiers, ranked by how hard each one would be to fake.

Solo Companies Hit $10 Million a Year — Even Before AI

Let’s go back to 2008. Markus Frind, based in Vancouver, Canada, was running the dating site Plenty of Fish entirely by himself. He’d built it in two weeks back in 2003, just to practice a new programming language, and through 2007 he had exactly zero employees. By 2008, revenue hit roughly $10 million — about ₩14 billion (~$10 million) — with margins above 50%. He was putting in about 10 hours a week. In 2015, Frind sold the company to Match Group for $575 million in cash. He’d never taken outside investment, so he held 100% of the equity himself.

There was no AI involved. What he used was the internet infrastructure of the era — ad networks, online payments, a handful of servers. So the phenomenon of a solo-founded company reaching $10 million in annual revenue existed well before AI. It’s been possible for 20 years. It was just very rare.

So what has AI actually changed? I’d argue it’s not scale — it’s frequency. According to Stripe’s data, the number of full-time American solopreneurs earning over $100,000 a year grew from roughly 2.5 million in the early 2010s to about 4 million by 2023. Cases like Frind’s — one person building and running a service alone — existed before. What AI does now is assist across multiple functions at once: writing code, drafting ad copy, handling customer support. I think this lowers the cost of starting a one-person company. But it doesn’t mean the entire long-term upward trend can be credited to AI.

And alongside these genuine cases, there have always been more fake ones. In 2023, the U.S. Federal Trade Commission (FTC) sued a company called Automators AI, which raised $22 million from investors by promising that “AI guarantees profits” through automated Amazon stores. In February 2024, the operators surrendered their assets and were permanently banned from the e-commerce coaching industry. What’s more telling is that the FTC keeps cracking down on these “automated income” schemes as if it were a running series — just last year, another company using the same playbook was sued again.

Across these 20 years of tangled real and fake cases, there’s one common thread: the evidence offered for success stories has usually been the founder’s own word and a screenshot of revenue. A screenshot alone can’t verify whether that revenue is real. This time, we can cross-check it against numbers compiled by payment companies and researchers.

Evidence Has Grades: Four Tiers Ordered by Cost of Manipulation

I sort the evidence behind solopreneur success stories into four tiers. There’s one criterion: how expensive it is to dress up the number in question. The further down the list you go, the higher the cost of faking it — and the more trustworthy the claim becomes.

Tier 1, self-reported. Screenshots, social media flexes, and run-rate¹ figures all belong here. Take Polsia, the marquee case in the Wall Street Journal piece. Founder Ben Broca says he started the company alone last December and built it up to 10,000 paying customers and $10 million in annualized run-rate revenue. Impressive. But that $10 million is simply recent performance multiplied by 12. The article itself notes that as his customer base grew, AI usage costs pushed him into the red, forcing him to switch to a free, open-source Chinese model. You need to look not just at annualized revenue but at whether there’s any profit left after AI costs. There’s one more thing worth noting: Polsia’s product is “a platform where AI runs your company for you,” and Broca himself says his own one-man operation is proof the product works. The protagonist of the success story is also the founder of the company selling that story. The fact that he raised $30 million in venture capital also sits oddly with the bootstrap-hero narrative. None of this means fraud. It just means a Tier 1 claim deserves exactly Tier 1 trust — no more.

Tier 2, payment data. This is where Stripe’s report from last June, “The Age of the Solopreneur,” belongs. The “2x, 3x” figures mentioned at the top of this piece come from this report — not a survey, but an actual tally of money that passed through Stripe’s payment network. That’s easier to verify than a screenshot someone posted themselves. Still, the limits of the statistics need checking too. The U.S. Census Bureau changed its counting methodology in 2022. Previously, one-person businesses whose revenue crossed a certain threshold were automatically reclassified as “employer firms.” Once that practice stopped, the number of solo businesses in the upper revenue brackets jumped statistically. Some of that surge isn’t new growth — it’s businesses being counted correctly for the first time. Stripe itself acknowledges this limitation in the report. And of course, this data only covers businesses that process payments through Stripe. Still, it’s clearly a number recorded in a payment company’s ledger, not a figure the subject reported about themselves.

Tier 3, third-party data. Numbers measured by someone with no stake in the outcome, neither the subject nor an interested party. The prime example is the working paper² “AI-Native Companies,” published last June by researchers at Harvard Business School and INSEAD. They cross-referenced roughly 50,000 startups with external workforce data and found that companies that built AI into their products had 25% fewer employees than peers in the same industry launched around the same time. In the broader sample, the gap was 12%. Yet valuations were comparable. Entry-level and manager headcount, in particular, ran about 15% lower each. The raw contrast is even starker: in the Y Combinator sample, AI startups averaged 13 employees versus 42 for the comparison group. The basic direction — “same value, fewer people” — holds up even when measured by outsiders. Administrative data points the same way. New business applications in the U.S. information sector rose nearly 45% in a year, even as hiring intentions in that sector fell faster than in any other industry.

Tier 4, price. Numbers where a third party conducted its own due diligence and paid its own money — that is, an acquisition price. Base44, an app builder started solo by Israeli developer Maor Shlomo, was acquired by Wix for $80 million in cash just six months after launch. An acquisition price is easier to verify than a founder’s own revenue claims. The acquirer runs due diligence and actually hands over cash, and if the judgment is wrong, it eats the loss. That said, an acquisition price doesn’t guarantee the business’s long-term success either. Believing a success story and betting $80 million on one are entirely different acts.

Once you’ve checked all four tiers, the last thing to look at is the distribution. According to Stripe’s figures, the top 10% of solo founders in 2025 posted first-six-month revenue 61 times the median. Even among people who started at the same time, the gap between the top tier and the middle is that wide. The people featured in the article sit at different points along this distribution too. Claire Vo, former Chief Product Officer at LaunchDarkly, is on track for seven-figure profit this year with a product used by 100,000 people, yet she says: “People overestimate how easy AI makes things, and underestimate the work it took me to get here.” On the other end is Samir Ahmad, who left nearly 20 years at Verizon to launch a solo AI consulting business. He used AI like a strategy and marketing advisor, but within a few months he shut the business down and has since gone back to working for a company. The top tier has genuinely moved up — but for people sitting at the median, things remain just as hard as before.

In a Country Where Success Stories Are the Product

I think Korea is exactly where this kind of grading-the-evidence exercise is most needed. Open any social media feed and you’re flooded with “certified proof” of “₩10 million (~$7,200) a month” and “zero-capital startups” — and a good chunk of those screens lead straight to a course payment page. A market has already formed where success stories circulate not as verified evidence but as products for sale. I actually worked out the conditions that make this kind of business possible in a paper I wrote in 2023, and it comes down to three things: information asymmetry³ that only the seller possesses, the fear of falling behind (FOMO), and an information-vulnerable audience with no means of verification. Put those three together, and the success story itself becomes the revenue model. This is exactly the skeleton of the “dream-selling business” we saw last month in the piece on the rooftop-room developer.

In Korea, we can turn to the official survey that combines the Statistical Business Register with sample questionnaires. According to the fact-finding survey by the Ministry of SMEs and Startups, there are 1,162,529 one-person creative businesses in the country — up 15.4% year over year. The growth itself is real. But look at the averages: annual revenue per business comes to ₩266,400,000 (~$190,000), with net income of ₩36,200,000. Divide that by twelve, and you’re looking at roughly ₩3,000,000 a month. The average age of a founder is 55.1, and the top reason for starting a business was “higher income” (40.0%). Averaged across the survey, the typical founder is in their fifties, has about 16 years of prior work experience before starting out, and nets around ₩3,000,000 a month. That’s why it’s hard to take the social-media stories of “twenty-somethings automating ₩10 million a month” as representative of one-person businesses as a whole. Wanting to earn more is, in itself, a healthy motivation — but that exact motivation is precisely the target success-peddling aims at. The gap between the success stories posted on social media and the statistical averages is the very thing success-peddling is selling.

Oswarld’s Lens

Honestly, I’m one of the people who falls into this statistic. I’ve been running a one-person consulting firm since 2020, which makes me one of the 1.16 million solo creative businesses in Korea. Writing about exploitation business models as a paper in 2023, I reached a conclusion from that vantage point: what keeps this kind of trade running isn’t lies so much as information asymmetry. The seller only needs to prepare one success story, but the buyer needs time and expertise to verify it. As long as the burden of verification stays this heavy on the buyer’s side, people selling success myths will keep showing up.

Exploitation Business: Leveraging Information AsymmetryThis paper investigates the “Exploitation Business” model, which capitalizes on information asymmetry to exploit vulnerable populations. It focuses on businesses targeting non-experts or fraudsters wharxiv.org

Drawing on my own experience building GTM strategies, this is where the significance of the tier classification lies. When there are numbers measured by someone else — payment data, acquisition prices — the time and expertise needed to verify a success story shrink. And as information asymmetry narrows, the profits that exploitation businesses skim off are the first thing to get cut. Just as disclosure requirements in finance shrank the room for market manipulators to operate, good data protects consumers simply by existing.

So my conclusion isn’t skepticism. I don’t think this trend is fake. Evidence at Tiers 2 through 4 — the kind that’s hard to fabricate — genuinely shows a shift in the revenue distribution of one-person businesses. But the verified real scale is always smaller than what the success myths claim, and there are people selling that gap. The next time you come across a one-person-business success story, ask just one question: what tier of evidence is this number? If someone is asking for payment while offering only Tier-1 evidence, what they’re selling isn’t a way to run a business — it’s a story.

Closing

The prophesied one-person unicorn doesn’t exist yet, but the number of million-dollar one-person businesses doubling is a fact confirmed by payment data. Evidence comes in tiers. Self-reported claims, payment data, third-party data, and acquisition price — in that order, manipulation gets harder. In a market where success stories are sold as products, these four tiers aren’t just an analytical tool — above all, they’re a standard to check before you spend your money.

Reader, at which tier did the most suspicious success story you’ve seen recently fall apart? Tell me in the comments where it broke down, and if you have a fifth tier of verification you’d add, I’d love to hear it.

---

💬 I’ll work the examples you share into the next issue. 📨 If you have a colleague worn out by success-story ads, pass this one along.

References & Further Reading

Primary sources

• The Wall Street Journal, “The Rise of Million-Dollar Companies With Just One Employee”, 2026. ··· This is where today’s piece started. The cases of Brocka, Bo, and Ahmad show both the high and low ends of the gauge nicely.
• Stripe Economics, “The Age of the Solopreneur”, 2026.6. Link ··· The original source for the “2x, 3x” figures and the 61x power law. You need to read all the way through the section on the Census measurement change to get the full picture.
• Kim, H. & Koning, R., “AI-Native Firms,” Harvard Business School Working Paper 26-090, 2026. Link ··· The original paper behind “25% fewer employees, same valuation.” The analysis of how entry-level hires and managers disappear first is the standout part.
• Ahn, K., “Exploitation Business: Leveraging Information Asymmetry,” arXiv:2310.09802, 2023 (revised 2024). Link ··· This is my own paper, which became the backbone of the Korea section. It maps out how information asymmetry and FOMO get assembled into an exploitation business. Please keep in mind it’s a preprint that hasn’t gone through peer review.
• FTC, “FTC Action Leads to Ban for Owners of Automators AI E-Commerce Money-Making Scheme,” 2024.2. Link ··· The court record of an “AI guaranteed income” scam. It shows what grows alongside a myth as the myth itself grows bigger.

Background

• Chafkin, M., “And the Money Comes Rolling In,” Inc., 2009.1. Link ··· A profile of Friend from 2008. You can see the “zero employees, ₩14 billion (~$10.8 million) a year” story firsthand, from an era before AI.
• Ministry of SMEs and Startups, “2025 Survey on the State of One-Person Creative Enterprises,” 2026.4. ··· The official statistics on Korea’s 1.16 million one-person businesses. Start here, not with your feed.

Related past issues worth reading

• Issue 164: Why Do Freelance Developers Moonlight as Delivery Riders?
• Issue 167: The Income Structure Behind a 30,000-Follower Creator
• Issue 176: Why Does the New Fed Chair Talk Like a Startup Founder?

---

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

📝 Glossary

Footnotes

1. Run rate: A figure obtained by multiplying the most recent month’s or quarter’s results by 12 or 4 to project it as annual revenue. It’s useful for showing growth speed, but if you pick the best possible period as your baseline, it inflates the number well beyond actual annual revenue. ↩

2. Working paper: A research draft that hasn’t yet passed a journal’s peer review. It lets you see the newest data as fast as possible, but you have to keep in mind that the conclusions may later be revised. ↩

3. Information asymmetry: A situation where only one party in a transaction knows the key information. The classic example is a used-car seller who alone knows the car’s defects — it’s cited as a textbook cause of market failure. ↩