One-Person Companies Are Surging—Here’s Why Most of Them Won’t Survive Long Enough to Matter

Alex Willen runs twelve e-commerce brands with no employees. He expects between $2.5 million and $3 million in revenue this year. Kristin Ginn left Microsoft in March. Seven months later she has roughly $250,000 received or contracted through year-end with no staff.

These stories circulate freely at dinners and conference panels. They make a tidy argument: the cost of running a company has fallen permanently. The problem is that the largest dataset backing this claim counts applications. It does not count companies that survived. A business application is a declaration of intent. Nothing more.

The real numbers are murkier than the testimonials suggest. Liya Palagashvili at George Mason University’s Mercatus Center tracked business filings in industries heavily exposed to AI. Applications showing no early hiring signals rose 26.8 percent from Q1 2024 to Q1 2026. That moved from about 222,000 to 281,000 per quarter. Applications that looked likely to become employers fell 6.4 percent. In construction and wholesale trade—the comparison group—non-employer applications slipped just 0.4 percent.

Palagashvili calls her findings preliminary and descriptive. She does not claim AI caused the rise. The Census measure flags filings without indicators of near-term payroll or hiring plans. It does not follow those businesses forward. They could hire later. They could close down. Nobody knows yet because longitudinal data matching those 281,000 quarterly applications against payroll records simply does not exist.

Data from firms that do see payroll points in the other direction entirely. Gusto tracked small businesses after they adopted AI. Headcount came in about 7 percent higher a year later compared to similar companies that knew about the tools but did not adopt them. Among firms with fewer than ten employees, adopters added staff about 10 percent faster than non-adopters. Both results can be true simultaneously. A tool that lets one person start a company also lets an existing five-person firm take on work it would previously have declined. The second effect shows up in payroll while the first shows up in filings.

The FreshBooks survey sits somewhere between these two signals. Seventy-nine percent of solopreneurs said they tried AI before hiring someone for a task they could not do themselves. The figure rose to 86 percent when microbusiness owners were included. Start cheap. Figure things out. Hire only when you have to. The instinct is rational. The base rate remains unknown.

The arithmetic of solo operation is worth understanding before declaring victory. Willen uses Anthropic’s Claude for bookkeeping, customer service, advertising and inventory. He relies on OpenAI’s image tools for work that once went to photographers. A product photography shoot still runs $2,000 to $5,000. A designer charges $300 to $1,000 for a set of listing images. Without AI, Willen estimates he would need one more full-time person plus contractors for photography, design, bookkeeping and ad management. His monthly spend on paid AI tools is under $60.

Ginn spends similarly little on software. Yet she rebuilt the Shopify store of Lone Trail, a hunting-gear brand she acquired. It had been running at roughly 1 percent of sales. In a recent 30-day stretch it accounted for about 30 percent of that brand’s revenue. She puts annual revenue from that effort at six figures. These are outcomes. Outcomes are not a base rate.

What neither model eliminates is dependence on distribution. Roughly 80 percent of Willen’s revenue arrives through Amazon. Meta and Shopify carry direct sales. Ginn has no staff. LinkedIn is among her largest sources of customers. The software bill is small. The platform bill is the business. You do not own your audience. You rent it.

Erik Brynjolfsson at Stanford reads this dependency as a feature rather than a bug. A longer tail of small producers raises the value of whatever aggregates them. Platforms sell the tools and supply the customers. His argument does not claim savings flow to big technology companies. Willen keeps some as profit. Ginn’s tool spend is trivial. The transfer is thinner than the rhetoric suggests.

The more consequential thesis concerns firm size itself. Brynjolfsson argued alongside Zoë Hitzig of the Anthropic Institute in a September paper that as AI codifies knowledge once held by individual workers and makes it available centrally, average firm size rises. Industry concentration increases. Local managerial discretion declines. They describe early evidence as broadly consistent with those patterns. Adam Branch’s experience illustrates the mechanism plainly.

Branch built Incentive Technology Group to about 500 employees. Winning more work meant hiring more people. Labour ran roughly 85 percent of operating spending. Technology sat near 5 percent. His current company, rhino.ai, has about 50 people. It does comparable modernisation work at three to four times the output per employee. Labour now sits near 40 percent. Technology including cloud, software and models accounts for nearly 25 percent.

Brynjolfsson draws a pattern from cases like that. A hollowing out of the middle. Very small teams doing more at one end. Very large organisations combining AI with scale at the other. Management layers built for internal coordination squeezed out. AWS surveyed 3,413 founders across 20 countries who run companies under five years old and use AI intensively. They reported average annual revenue growth of 156 percent against 65 percent for startups generally. More than half exceeded $400,000 of revenue per employee. Ruba Borno, who runs AWS’s specialists and partners group, called the change scaling with technology instead of headcount.

Surveys of that kind select for survivors. A solo operator with a failed storefront does not appear in a sample of active founders. The $400,000 per employee figure is an average drawn from companies still trading when the survey landed. The honest reading until longitudinal data emerges is straightforward. The cost of trying has fallen. The cost of failing has not.

Twenty-eighty-one thousand quarterly applications in AI-exposed industries can eventually be matched against payroll records. In three or five years we will see how many became employers. How many stayed at one person. How many disappeared. Nobody has published that yet. Until then Ginn’s $250,000 and Willen’s $3 million remain outcomes. Not a base rate. The filings analysts are counting describe an intention to find out. The finding out has barely started.

The move is clear: audit your own dependency stack before celebrating automation. Map every dollar flowing to platform fees, ad buys and distribution intermediaries. Distinguish between tools that compress execution and tools that substitute for market access. Build at least one revenue channel you control directly or negotiate explicitly. Solo operations survive on distribution luck. They compound on ownership.