Harvard Business School, INSEAD study finds AI-native startups 25% smaller than traditional firms

The startup playbook has always assumed growth means hiring. More customers, more employees, bigger org chart. A new working paper from researchers at Harvard Business School and INSEAD suggests that for AI-native companies, that equation is breaking down.

The study, released June 9, 2026, finds that startups embedding AI directly into their products employ roughly 25% fewer workers than comparable non-AI startups, while maintaining similar company valuations. In plain terms: same value, fewer people on payroll.

What the research actually found

Researchers Hyunjin Kim of INSEAD and Rembrand Koning of Harvard Business School analyzed Y Combinator batches from Winter 2020 through Fall 2024, alongside U.S. venture-backed startups that received their first funding between 2020 and 2024.

The workforce gap is significant. Using log specifications, the coefficients land around -0.28, translating to approximately 25% fewer employees at AI-native firms relative to peers in the same industry cohort.

It is not just headcount that looks different. The internal structure of these companies is flatter, roughly half a seniority level less deep than traditional startups. Think fewer middle layers between junior staff and the executive team.

The composition of that smaller workforce also shifts noticeably. AI-native startups carry about 13% more engineers as a share of total employees. At the same time, they employ approximately 15% fewer entry-level workers and managers.

The researchers draw a careful distinction here. The operational advantage comes from embedding AI into the product itself, not from bolting AI tools onto existing workflows. A company using ChatGPT to write marketing copy is not the same animal as a company whose core product is built on AI capabilities from day one. The former gets incremental efficiency gains. The latter restructures what the company needs to exist at all.

Why this matters beyond the startup world

The valuation parity finding is the detail worth sitting with. If an AI-native firm and a traditional startup in the same sector command comparable valuations, but the AI-native firm gets there with a quarter fewer employees, the value-per-employee ratio at these companies is materially higher.

For investors, that changes the math on what a promising early-stage company looks like. Headcount growth has historically served as a proxy signal for business momentum. If AI-native firms are structurally lean by design rather than by distress, that proxy becomes unreliable.

The decline in demand for entry-level roles at AI-native firms is not an accident of the sample. It is a feature of the operational model. Junior employees often handle tasks that, in an AI-native company, the product itself handles.

The reduced manager layer compounds this. Flatter hierarchies mean fewer rungs for ambitious junior employees to climb, and fewer management positions being created as these companies scale.

Investors scouting early-stage deals may want to recalibrate what healthy looks like. A 30-person AI-native startup with strong revenue metrics and a high engineer-to-total-staff ratio could represent a more capital-efficient business than a 40-person traditional startup at the same valuation.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.