AI-first company

Study

Startups most exposed to AI cut staff, mostly juniors, and grew faster

Startups whose work was fully exposed to AI cut employment by 8% within two quarters of ChatGPT's release, mostly in junior and implementation roles. Their code contributions rose 64%, and they became more likely to rank among the startups with the highest sales per employee.

By
Last reviewed
Type of study
Data analysis
Published
2026
Corrections
Report an error

At a glance

Source
Gupta, Qian, Simintzi and Sun (2026). Generative AI and Entrepreneurship. NBER conference paper, version of 14 April 2026. conference.nber.org · Not peer-reviewed: other researchers have not yet checked it for a journal.
How it was done
The authors measured how much of each startup's work AI could do, from its mix of jobs just before ChatGPT's release. They then compared more and less exposed startups before and after the release, using workforce, code and sales records. Data analysis. It builds a measure from existing data, such as job and task records.
Who took part
94,789 US startups founded between 2018 and 2021 and listed in Crunchbase or PitchBook. On average they employ 11 workers, have raised $3 million and generate about $700,000 a year in revenue.
Where
United States; young, fast-growing firms of the kind that seek venture capital, in all sectors. Information technology is the largest sector, at 36.8% of firms.

“Overall, incumbent contraction was offset by new firm formation, leaving aggregate employment unchanged but shifting composition to senior roles.”

What it found

  • Fewer staff. Startups whose tasks were fully exposed to AI employed 8% fewer people than before ChatGPT’s release. The fall came within two quarters and lasted to the end of the data in 2025. Roughly one-third of the tasks in the sample were exposed, so the typical startup’s fall was smaller.
  • Juniors went first. Most of the cut fell on less senior staff and on people in implementation roles. The share of senior staff rose.
  • Leavers paid for it. About half of the people who left waited about six months for a new job. Many took lower-paid jobs that AI could do less of.
  • More output. Code contributions on GitHub at fully exposed startups rose 64%. These startups became about 20% more likely to rank among the top startups by sales per employee.
  • More funding. They were 21% more likely to raise venture capital than before.
  • More, smaller startups. In markets more exposed to AI, the number of new startups rose 6% and average startup size fell 4%. Total employment stayed about the same.

When the result holds

The startups sell digital output and are funded to grow. For them, AI took over work that juniors used to do.

Other evidence on small firms points the other way. In Gusto’s payroll records, US firms with fewer than 10 employees that adopted AI hired more people over the next year, mostly hands-on, customer-facing staff (Tremper and Bowen, Gusto, 2026). The two sets of firms differ in what they sell, and the two studies measure AI differently.

Limits

  • A working paper, not peer-reviewed.
  • The study measures how much of the work AI could do, not whether the startups used AI.
  • Venture-backed startups are not typical small firms such as shops, trades or local services.
  • The sales and code measures cover only some of the firms.

Practices that use this study

This study supports: