Cheyenne, Wyoming produced new businesses per resident at twice the rate of New York City this year, according to payment processor Stripe. In 2022, the two cities were tied. That single comparison captures a broader pattern in Stripe’s sign-up data: companies founded since 2024, the period Stripe treats as the start of the AI era, are launching further from the traditional business hubs than any prior cohort the company has tracked.
Roughly 40 percent of this year’s new businesses on Stripe are located in metro areas with fewer than a million residents, a share that has grown by ten percentage points over four years. Smaller and mid-sized metros, Hinesville, Georgia and Fayetteville, North Carolina among them, now register some of the densest new-business activity in the country. New York, Los Angeles, and San Francisco are all forming businesses at below-average rates relative to their populations.
The dispersal is not confined to which metro a founder picks. Inside major cities, formation is drifting from urban cores to outer suburbs. In Houston, the outer-suburb share of new businesses climbed to 37 percent this year, a four-point jump from 2023. Austin, Tampa, Orlando, and Dallas show comparable three-to-four-point shifts. Roughly 80 percent of sign-ups in major metros now sit outside city centers and the densest inner ring, tracking population distribution rather than concentrating near it.
Stripe is careful about what caused this, and the discipline is worth preserving rather than smoothing over. The company’s own analysis states plainly that business formation does not correlate with work-from-home intensity, whether compared across metro areas or within them. That is a specific, testable claim, not a hedge. It means the popular shorthand, that AI is simply extending a remote-work migration that began during COVID-19, does not hold up against Stripe’s own numbers. The trend line accelerated again starting in 2024, alongside AI’s spread through small businesses, but acceleration alongside a technology is not the same as acceleration because of it. Stripe sizes the current shift at roughly a third of the pandemic-driven exodus from big cities, and it stops short of assigning AI a causal role.
The detail that sharpens the story sits in Stripe’s own data, one paragraph past the headline finding. Split new companies by proximity to the technological frontier and the dispersal pattern reverses. Nearly 80 percent of AI labs sit in San Francisco and nowhere else. Half of AI product companies sit in San Francisco or a top-ten metro. Only the third category, businesses that use AI tools but do not build them, spreads out the way the aggregate numbers suggest.
That split is the actual news here, and it is one the source data supports without needing to guess at intent: the companies producing the tools that let a solo founder in Cheyenne skip hiring a designer or a lawyer are themselves clustering harder around each other, not less. Agglomeration, the economic case for putting frontier firms next to other frontier firms, appears to be strengthening exactly where AI is made even as it weakens for everyone using what gets made there.
For operators, the practical read is narrow but useful. If your business consumes AI capability rather than builds it, geography is now close to irrelevant to where you can start. If you are trying to build at the frontier, competing with labs and product companies for talent and capital, San Francisco’s density advantage has not eroded. Founders should stop treating “AI makes location optional” as a universal claim and start asking which category their company actually falls into before picking a zip code.
Stripe Economics published this analysis, written by Marisa Rama, on August 27, 2026.
