The headcount was the vanity

A Series A used to buy people. In 2020 the median one landed at about $10 million and, spread across the team it paid for, worked out to roughly $160,000 a head. By 2025 the median round had grown to $15 million and the figure per employee had more than doubled, to north of $320,000, according to an October analysis of hiring and funding data by Revelio Labs. The money got bigger. The teams it hired got smaller. Over the same stretch the median Series A company shrank from 57 employees to 44.

None of this happened in a vacuum. The efficiency era that began with the 2022 correction never really ended: 2024 alone saw more than 150,000 tech job cuts across 549 companies, by the count of the tracker Layoffs.fyi. What started as retrenchment hardened into a preference. Founders, and the investors backing them, stopped treating a growing payroll as evidence of progress and started treating it as a cost to be justified.

The math turned over

The clearest sign is a number that used to barely move. The average public SaaS company generates about $300,000 of revenue per employee, and for years a figure in that neighborhood was what a healthy software business looked like. AI-native startups have detached from it entirely. Writing in Forbes in March, Paul Baier put their range at $2 million to $4 million a head — an order of magnitude above the benchmark, and above incumbents that were themselves considered lean.

The anecdotes are almost unfair. Midjourney, the image generator, reached around $200 million in revenue with about eleven people — roughly $18 million each. Baier notes that the top twenty AI-agent startups tracked by CB Insights now post revenue-per-employee figures above Microsoft’s $1.8 million and Meta’s $2.2 million, firms whose whole reputation rested on doing more with fewer engineers. The coding tool Cursor, built by Anysphere, crossed $500 million in annual recurring revenue in mid-2025 with revenue reportedly doubling about every two months — a curve that no longer tracks the hiring plan at all.

Headcount as a status symbol

For a long time the size of the team was the pitch. “We’re a hundred and fifty people now” did work in a room that the revenue figure sometimes couldn’t; a big org chart signaled traction, ambition, the confidence to spend. Headcount was the vanity metric nobody called a vanity metric, because it looked so much like the real thing. You could grow it on purpose, put it on a slide, and let a valuation follow.

Altman’s bet is closer to a party trick than a forecast — the one-person unicorn is a threshold, not a plan. But the direction it points is the one the data already shows. When capital buys half as many people, and a lean team can out-earn a large one per head, the incentive to hire ahead of need thins out, and with it the reason to brag about the number.

Leaner is not the same as free, and the efficiency story flatters its winners. A small team with no slack is a fragile one: no bench when someone leaves, no spare capacity when a launch slips, and the same automation that lets nine people act like ninety lets a competitor do it right back. The startups clearing nine figures on a rounding error of staff are outliers, and the median company still runs on people doing unglamorous work. The change isn’t that everyone got lean. It’s that the ones who did stopped apologizing for it.

So the boast has inverted. Where a founder once led with how many people the company employed, the sharper flex now is how few — how much revenue each seat carries, how long the runway stretches without another raise. The headcount was the vanity all along. What replaced it is a smaller number that means more, and is harder to fake.