On average, it takes about 28 inbound inventors to generate one new venture-backed startup in a region.
That’s not a vibe check or a founder anecdote. It’s a causal estimate from large-scale patent and financing data, and it puts a hard number on something most of us only feel: when serious technical talent moves into your ecosystem, startup formation actually moves with it.
The research behind this goes a lot deeper than “smart people move in, startups happen.” It uses millions of inventor records to separate cause from correlation and shows that inventor migration doesn’t just add more startups — it nudges the whole quality curve up.
How Patent Trails Turn Into a Map of Inventor Migration
The core data source is simple: patents. Every U.S. patent lists inventors and their locations at the time of filing. By following those addresses over time, researchers can see when inventors move from one county to another.
In this study, they track inventor moves across U.S. counties from 2000 to 2016. That yields over 21 million inventor-location observations tied to more than 180,000 individual inventors, giving a granular view of how technical human capital flows between regions.
Those moves are then matched to where and when new venture-backed startups appear. If a county suddenly sees a spike in inbound inventors and, a bit later, a spike in VC-backed startups, that’s a signal — but on its own, it’s still just correlation.
Why This Isn’t Just Correlation: The Shift-Share Trick
To get at causality, the paper leans on a shift-share instrumental variable (SSIV) approach. In plain English: it uses external shocks to inventor demand in specific technologies, combined with each region’s historical specialization, to predict where inventors will move next.
Think of it this way. If AI patenting booms nationally, areas that were already strong in AI patents are more likely to attract AI inventors, independent of any local startup policy or new accelerator. That predicted inflow becomes the “instrument” — a quasi-random push of inventors into certain regions.
By comparing startup outcomes driven by these instrumented inventor inflows — rather than just raw observed inflows — the researchers can argue that the extra startups are caused by inventor migration, not by some unobserved local factor. The formal details are in the NBER working paper that underpins this analysis.
There are limits. Patent data has lags: an inventor might move, start working at a startup, and only later file a patent that shows their new location. That timing noise makes the estimates conservative, if anything, but it’s still a constraint you should keep in mind when thinking about “real-time” talent strategies.
What 28 Inventors Actually Do Inside the Startup Pipeline
The headline number — 28 inbound inventors per additional VC-backed startup — is only part of the story. These inventors don’t just sit in labs; they show up across the startup stack.
Inbound inventors are directly involved in more than 1 in 13 new venture-backed startups, either as founders, executives, or named inventors on the startup’s patents. That’s a surprisingly high share for a relatively small slice of the overall workforce.
When they plug into startups, those companies look different. The research finds that startups with inbound inventors involved tend to raise more capital, attract stronger venture firms, and generate higher financial returns. In other words, you’re not just getting more companies — you’re getting a better distribution of outcomes.
This is the dual effect of specialized human capital: it increases startup quantity and shifts the quality curve. The presence of experienced inventors raises the ceiling on what a local startup can credibly attempt — deeper tech, more defensible IP, more complex products — which in turn attracts different investors and partners.
Why the Same Inventor Move Plays Very Differently in Different Places
The 28-inventors-per-startup figure is an average, and it hides big regional differences. The strongest effects show up in established startup hubs: think Bay Area, Boston, New York, and a handful of other dense tech metros.
In those places, inbound inventors are landing in ecosystems that already have VC, experienced operators, legal and banking infrastructure, and acquirers. Add a few more high-skill people and the system knows how to turn that into companies.
In rural counties or low-entrepreneurship regions, the same inflow of inventors doesn’t translate into the same startup bump. The local “conversion rate” from talent to venture-backed companies is much lower, and in some places, statistically negligible.
That’s a key caveat: the SSIV estimates are local average treatment effects. They tell you what happens when you push more inventors into places that already look like the sample’s active regions. If you’re in a county with almost no startup activity, you shouldn’t expect the 28:1 ratio to hold.
There are also spillovers that aren’t all positive. As more inventors and VC dollars concentrate in high-tech sectors, lower-tech local industries can lose out on capital and attention. That reallocation can be good for aggregate innovation but painful for legacy employers in the same geography.
Even with robustness checks, spatial correlation is hard to fully kill. Nearby counties share labor markets, investors, and institutions, so some of what looks like a “local” inventor effect may be driven by broader regional dynamics. For founders, the practical read is: think in terms of metro ecosystems, not county lines.
The Numbers That Really Matter for Founders and Ecosystem Builders
Here are the core quantitative results that define this inventor migration–startup growth story:
- Roughly 28 inbound inventors are associated with one additional venture-backed startup in a county, using the causal SSIV estimates.
- Inbound inventors are directly involved — as founders, executives, or patent inventors — in over 1 in 13 new VC-backed startups.
- When patent citations in a metro area double, inventor moves into startups in that metro rise by about 23%, showing how perceived technological opportunity pulls talent.
- The underlying dataset covers over 21 million inventor-location observations from more than 180,000 inventors between 2000 and 2016.
For founders, this reframes “talent strategy” from a soft HR topic into a lever with measurable ecosystem effects. If you’re building in a hub, every senior inventor you convince to relocate or spin out of a big company doesn’t just help your company — they change the odds for the next wave of startups around you.
For people running ecosystems, it suggests that targeted efforts to attract inventors — especially in frontier domains like AI, where ecosystem dynamics are already winner-take-most — can have compounding returns. The recent Cornell writeup on “talent sparks” in startup regions is basically this logic in narrative form.
And if you’re an inventor thinking about moving: the data says your choice of where to land doesn’t just change your career. It nudges an entire local startup distribution — both in how many companies get formed and how ambitious they can afford to be.
