# More firms start alone in the sectors AI touches most

> Census data analysed by the Mercatus Center show new business registrations without hiring signals rising 26.8% in AI-exposed sectors since early 2024, while those that look like future employers fell 6.4%. The author says the finding is descriptive, not causal, and the debate over what it means has only started.

- URL: https://ai-news-daily.xyz/posts/more-firms-start-alone-in-the-sectors-ai-touches-most/
- Published: 2026-11-08
- Tags: business, research
- Author: AI-generated (this text was generated by AI; see https://ai-news-daily.xyz/about/#ai-disclosure)
- Publisher: [AI News Daily](https://ai-news-daily.xyz/)

Suppose a management analyst in her forties leaves a consulting firm in January. She has no investors and no plan for a first hire. She registers a company, opens a Claude subscription for the research and the slide drafts, hands bookkeeping to an accounting app and tells her old clients she is now available by the project. This woman is invented, a stand-in for a pattern, and the pattern is what a new analysis of federal data is trying to measure.

The analysis comes from Liya Palagashvili, director of the Labor Policy Project at the Mercatus Center at George Mason University. Her paper, published on 17 July 2026, looks at business applications filed with the Census Bureau. Between the first quarter of 2024 and the first quarter of 2026, applications that carry no early sign of hiring rose 26.8% in AI-exposed sectors, from about 222,000 to about 281,000 a quarter. Applications that look like future employers fell 6.4%. In a comparison group of construction and wholesale trade, the solo-type count moved by minus 0.4%. CNBC reported the result on 7 November.

## How the number was built

The Census Bureau publishes Business Formation Statistics, counts of applications for employer identification numbers. It also flags a subset as high-propensity: applications whose traits, such as a corporate structure, planned wages or hiring intent, have historically gone with firms that later employ people. Palagashvili subtracts that subset from the total and calls the remainder "solo-type" formation. She is open that this is a proxy she constructed and not an official Census series.

She defines the AI-exposed sectors as professional services, information, education, and finance and insurance, chosen for their high AI use in the Census Business Trends and Outlook Survey. Construction and wholesale trade serve as the control. She left out real estate, manufacturing and retail because of other shocks in those years. Total applications in the exposed sectors rose 18.0% over the period.

An event study across seven sectors puts the gap at about 16% more solo-type growth in the exposed sectors by 2025. The author notes that with only seven sectors the standard errors are likely understated, and that the 2024 coefficient alone is positive but not significant.

## What survey data add

A second data set, the Current Population Survey, tells a similar story from the worker side. From a 2022 to 2023 baseline to 2025, solo self-employment rose 7.9% in AI-exposed industries and fell 2.1% in the comparison group. In the ten occupations the Felten exposure index rates as most exposed to AI, solo self-employment rose 20%; in the ten least exposed it barely changed.

Management analysts are the paper's case study. The occupation grew from about 1.0 million to 1.12 million workers, and the share working solo climbed from 5.6% to 7.0%, near a decade high. Palagashvili describes the occupation as reorganising, not shrinking.

## The caveats the author states

She calls the evidence "preliminary, descriptive, and not causal". The exposure index predates the latest generative models. The first quarter of 2026 covers only January to March, the quarterly data are not seasonally adjusted, and other shocks specific to a sector could be driving the split. Palagashvili told CNBC the findings do not show that AI caused the rise.

That restraint matters, because the story invites a tidy reading in which cheap software turns every professional into a one-person company. The data show where registrations went. They do not show why, and they say nothing yet about how many of those solo firms survive or earn a living.

## Accounts that point both ways

CNBC set the numbers beside business owners. Alex Willen runs twelve e-commerce brands with no employees and expects $2.5 million to $3 million in revenue this year. He told the outlet that product photography would cost $2,000 to $5,000 and listing designers $300 to $1,000 a set; Claude and ChatGPT helped him rebuild the Shopify site of a hunting-gear brand he bought. About 80% of his revenue still comes through Amazon. Kristin Ginn left Microsoft in March and says she has about $250,000 in business received or contracted through year-end, on under $60 a month of paid AI tools and no regular contractors.

Vendor surveys complicate the picture. Gusto, the payroll software firm, found that small businesses using AI had about 7% higher headcount after a year than comparable non-adopters, and about 10% higher among firms under ten people, as CNBC relays it. A survey by Amazon Web Services of 3,413 founders found that AI-advanced young startups grew revenue 156% a year on average against 65% for startups overall. Both come from companies that sell to small businesses, and neither measures the same thing as the Census registrations. They do show that firms using AI are not simply declining to hire.

## An old question about the firm

The economics behind the debate are older than the models. In 1937 Ronald Coase asked why firms exist at all, and answered that using the market has costs: finding people, agreeing terms, checking the work. A firm hires when those costs are higher than managing staff. Palagashvili's paper frames AI as a possible drop in some of those costs, which would make contracting out, or doing work alone, more attractive for tasks that once needed an employee.

Erik Brynjolfsson of Stanford and Zoë Hitzig, in a chapter published in September, argue the opposite pressure also exists. AI can codify local knowledge and make central control easier, so average firm size and concentration could grow. Brynjolfsson told CNBC he expects a hollowing out of midsize firms, with solo operators and very large platforms gaining. Adam Branch, who founded the software company rhino.ai, estimates labour was about 85% of operating spending at his earlier 500-person firm and about 40% at his current one of roughly 50.

Palagashvili's policy conclusion follows from the first reading. Unemployment insurance and employer benefits assume a W-2 job, so she proposes portable benefits that follow a worker between contracts and self-employment.

## What to watch

Two things will test the finding. One is the next few quarters of Census data, which show whether the divergence widens, holds or fades. The other is survival: a registration is a form filed, not a business that lasts.

Back in the supposed analyst's first year, a client asks whether she can take on a second engagement in March and a third in June. She can do the work. What she cannot yet do is answer who covers her if she falls ill, and nothing in the Census series can tell her how many others are asking the same.

## References

- Brynjolfsson, E., & Hitzig, Z. (2026). *AI's Use of Knowledge in Society*. In *The Economics of Transformative AI*, NBER. https://www.nber.org/books-and-chapters/economics-transformative-ai/ais-use-knowledge-society
- Booth, B. (2026). *America's new no-employee small business owners start hot in AI era*. CNBC. https://www.cnbc.com/2026/11/07/americas-new-no-employee-small-business-owners-start-hot-in-ai-era.html
- Coase, R. H. (1937). *The Nature of the Firm*. Economica, 4(16). https://doi.org/10.1111/j.1468-0335.1937.tb00002.x
- Palagashvili, L. (2026). *Artificial Intelligence (AI) and the Rise of Independent Work: Early Evidence from Solo Entrepreneurship*. Mercatus Center at George Mason University. https://www.mercatus.org/research/research-papers/artificial-intelligence-ai-and-rise-independent-work-early-evidence-solo
- U.S. Census Bureau. *Business Formation Statistics*. https://www.census.gov/econ/bfs/index.html