Seed-stage startups averaged 6.4 employees in 2022. By 2024, that number dropped below 3.5. That's not a recession story. It's an AI story.
Laurie Voss has pulled together a compelling set of numbers showing that AI isn't just helping companies do more with less - it's fundamentally reshaping how many people a startup needs to function. The data is stark: AI-native startups generate $3.48 million in revenue per employee, compared to $580,000 for traditional SaaS companies. That's a 6x gap. At $10 million in annual revenue, an AI startup needs 15 to 20 employees where a traditional company would need 50 to 70.
The Money Is Flowing, the Jobs Are Not
February 2026 set a record: $189 billion flowed into startups in 28 days. AI-related companies captured roughly half of all venture funding in 2025, totaling $211 billion. But here's the uncomfortable part - three companies alone (OpenAI, Anthropic, and Waymo) absorbed 83% of February's funding.
Monthly startup hiring has declined over 50% between January 2022 and January 2024. Over 126,000 tech workers were laid off in 2025. Series A median headcount fell from 57 in 2020 to 47 today. Block cut 40% of its workforce, citing AI efficiency gains.
The pattern is consistent across every funding stage: companies are getting funded at higher valuations with smaller teams.
Where Are the New Roles?
Every previous wave of automation - the spreadsheet replacing accountants, the internet replacing travel agents - eventually created more jobs than it destroyed. The optimistic view is that AI will do the same: more companies will get started, more products will ship, and new categories of work will emerge.
But the data doesn't show that happening yet. The number of new startups hasn't increased to offset the smaller team sizes. The venture funding is concentrating, not spreading. And the laid-off workers aren't being absorbed by a boom in AI-adjacent roles.
Voss's argument isn't that AI will permanently destroy jobs. It's that right now, in early 2026, compute is substituting for labor and the compensating job creation hasn't arrived. That's a real problem for anyone job-hunting in tech today.
For teams already using AI tools, this validates what many already feel: a three-person team with good AI tooling can genuinely ship what used to require ten people. The question nobody can answer yet is whether that means we'll get three times as many teams or just the same number of teams with fewer people.