Mega AI seed rounds aren't venture gold—data shows 1% of $100M+ first rounds hit 10x returns

What happened

Analysis of roughly 200 publicly disclosed $100 million（約160億円）-plus first rounds over 15 years found that only 20% had recorded exits, and of those, only a few delivered 10x or better returns for first-round investors—roughly 1% overall. The finding challenges the narrative that headline-grabbing mega-seed rounds like Yann LeCun's $1 billion（約1600億円） and Project Prometheus's $6.2 billion（約9900億円） represent a new venture paradigm.

Why it matters

High entry prices in mega-seed rounds leave less room for upside to compound, hurting investor returns even when the underlying company succeeds. Historical winners like Google and Uber generated 300x–5,000x returns for early investors partly because those investors entered at much lower valuations. Even AI winners often started small: Cursor raised less than $10 million（約16億円）, ElevenLabs $2 million（約3.2億円）, Cohere $5 million（約8億円）—all now valued above $5 billion（約8000億円）. The data suggests capital intensity alone does not create venture-scale returns.

What to watch

While OpenAI and Anthropic are projected to improve the dataset when they exit—first-round investors there are looking at 30–40x returns at projected IPO valuations—that outcome still lags historical standards. The playbook that has worked across prior technology waves remains buying meaningful ownership in capital-efficient companies at prices that leave room for upside, not chasing mega-round exceptions.