Two pictures of the same economy
One picture: AI tools let a single founder do what used to require a co-founder and two hires, and a wave of ultra-lean, high-revenue one-person companies is the result. The other picture: AI is quietly hollowing out the entry-level jobs that used to teach people the skills they’d need to found anything at all. Both pictures are drawing on real 2025-2026 data. Neither is the whole economy. This article tries to hold them at the same time.
“The sky is not falling, but it is slowly lowering”
The most useful framing predates the current hype cycle. MIT’s Task Force on the Work of the Future, co-chaired by economists David Autor and David Mindell, released its final report in November 2020 after two years studying automation’s actual effects on American workers. Autor’s summary line has aged well: “The sky is not falling, but it is slowly lowering.” The task force’s broader finding was that roughly 63% of the jobs held in 2018 did not exist in 1940 — technological change constantly destroys and creates categories of work — and that the more urgent problem was not robots stealing jobs wholesale, but policy failing to let productivity gains reach workers: a stagnant minimum wage, weak unemployment insurance, and eroded collective bargaining power.
That framing matters for entrepreneurship specifically. If the core problem is distributional rather than a shortage of work to do, then AI-enabled solo entrepreneurship is one plausible release valve — a way for displaced or underpaid expertise to capture value directly instead of waiting for an employer to pass along productivity gains. Whether it functions that way at scale is an empirical question, not a foregone conclusion.
What the solopreneur numbers actually say
Stripe’s in-house research arm, Stripe Economics, published “The Age of the Solopreneur” in June 2026, authored by Ernie Tedeschi, Marisa Rama, and Chris Cruickshank. It found roughly 4 million Americans earned primary income as solopreneurs generating over $100,000 a year in 2023, up from the mid-2 millions in the early 2010s. The share crossing $1 million in income roughly doubled between 2023 and 2025. Delaware incorporations — a common proxy for serious startup formation — were up about 40% year-over-year since early 2025, and the 2025 cohort of new Stripe-processing businesses reached $1 million in revenue at roughly a 30% higher rate than the 2023 cohort.
Read that with one caveat firmly attached: Stripe Economics is a genuinely capable research shop, but it is Stripe’s own payments-processing data, published by a company whose business grows when more people incorporate solo ventures and process payments through Stripe. That is not a reason to dismiss the numbers — it is a reason to want independent corroboration before treating them as neutral macroeconomic fact.
That corroboration exists, and it comes from a source with no stake in the narrative: the U.S. Census Bureau’s Business Formation Statistics. The Bureau’s own annual release — which folded in full-year 2025 data in June 2026 — recorded 5.62 million business applications filed in 2025, up from 5.2 million in 2024 and well above the roughly 3.47 million annual average since 2005. Applications in 2026 have continued running meaningfully above the same period in 2025. Government business-registration data cannot distinguish “founded because of AI tools” from “founded for any other reason,” but it independently confirms that something real is happening in business formation, at a scale too large to be an artifact of one payments company’s customer base.
What 2025-2026 labor data says about actual disruption
Two of the most careful ongoing research efforts on AI and employment both land on a similar headline, with one important exception.
The Budget Lab at Yale has run a rolling analysis of AI’s labor-market footprint since shortly after ChatGPT’s public release. Its October 2025 update concluded that “the broader labor market has not experienced a discernible disruption since ChatGPT’s release 33 months ago, undercutting fears that AI automation is currently eroding the demand for cognitive labor across the economy.” That is a strong, if unglamorous, finding: no broad wave of AI-driven layoffs shows up in the aggregate data, over two and a half years in.
The exception is where Stanford’s Digital Economy Lab has focused its own rolling analysis. Economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, in their periodically updated “Canaries in the Coal Mine?” study using ADP payroll data through mid-2026, find no evidence of widespread displacement either — but they do find that employment of workers aged 22-25 in the most AI-exposed occupations now sits about 19% below where it would be had it kept pace with less-exposed peers. There is no comparable gap for experienced workers. Critically, the effect runs almost entirely through reduced hiring, not increased firing: employers are not laying off junior staff because of AI, they simply are not hiring as many of them in the first place.
Put these two findings together and a coherent, narrower story emerges: AI is not currently producing a broad jobs crisis, but it may already be narrowing the entry ramp into some white-collar career tracks — the same ramp that historically produced experienced professionals who could later strike out on their own. That is a genuinely different concern from “AI is eliminating jobs,” and it deserves to be treated as such rather than folded into either the boom or bust narrative.
Two ideas that sound plausible and are not yet evidence
Two claims circulate constantly in essays and founder Twitter threads about AI-era entrepreneurship, and neither has the institutional or peer-reviewed backing that the figures above carry.
The first is that “taste” — curatorial judgment about what’s worth making — becomes the durable moat once execution is cheap. It’s an intuitive argument, and plausible as far as it goes, but it is commentary, not research; no rigorous study establishes that curatorial judgment resists AI-assisted commoditization any better than the execution skills it’s contrasted with.
The second is that local, relationship- and trust-based businesses are inherently AI-resistant. Some argue this on the theory that trust is built face-to-face and can’t be faked at scale. But the evidence available actually cuts in the opposite direction at least partially: AI-driven recommendation and search are rapidly becoming the first step in how people find local services at all, meaning the AI layer is inserting itself upstream of the “trust” a local business relies on, not leaving it untouched. Treat both ideas as hypotheses worth watching, not settled findings — this manual’s sourcing standard requires that distinction to stay visible.
Where that leaves a household weighing the decision
Structural business-formation growth and a softening entry-level hiring market for young AI-exposed workers are not contradictory; they may be two faces of the same shift. Fewer employees per venture is consistent with both more solo founders and fewer junior hires. Nothing here justifies “quit your job and become a solopreneur” as generic advice, and nothing justifies “don’t bother, AI already ate this” either. The next article in this chapter turns these findings into concrete household planning guidance, including what someone in an AI-exposed early-career track can actually do about the hiring gap Stanford’s researchers documented.