In February, US startups raised $174 billion. In March, they raised $13 billion. That is a 93 percent decline in thirty days, and the explanation is almost embarrassingly simple: February's total was fiction, and March's total is closer to reality. [1]

Three rounds accounted for nearly all of February's haul. OpenAI closed a $110 billion raise. Anthropic secured $30 billion. Waymo added $16 billion. Remove those three transactions and February's total drops to roughly $18 billion — still higher than March, but in the same neighborhood. The headline collapse is not a collapse at all. It is the market returning to its actual size after three companies briefly made it look ten times larger than it is. [2]

The Crunchbase data tells a more granular story. Late-stage funding — Series C and beyond — has essentially frozen. The mega-rounds that defined 2024 and early 2025 have concentrated into a handful of AI infrastructure companies, leaving the rest of the late-stage market starved for capital. Investors who once spread bets across dozens of growth-stage companies are now writing fewer, larger checks to fewer, larger companies.

Seed funding, by contrast, has remained roughly flat. Early-stage investors are still writing $2 million to $5 million checks into new companies at roughly the same pace they were a year ago. The bifurcation is striking: the venture market has split into two economies. In one, a small number of AI companies absorb tens of billions. In the other, everything else competes for scraps.

European funding provides a useful counterpoint. March 2026 is actually the strongest month for European startups this year, with several mid-stage rounds closing in Berlin, London, and Paris. The capital that is retreating from US late-stage deals has not disappeared from the global system — it has shifted geography and stage.

Forbes greeted the data with the headline "Welcome to the Value Creation Era," arguing that the end of cheap capital forces startups to build real businesses rather than subsidize growth with investor money. There is a version of this argument that is correct: companies that can generate revenue will survive, and those that cannot will not. But there is also a version that is euphemism — a way of saying that the money is gone and calling it discipline.

The question the data cannot answer is whether the AI megarounds represent genuine value or the last act of a cycle that has confused scale with substance. OpenAI's $110 billion raise valued the company at roughly $300 billion. It has never turned a profit. Whether that is visionary or delusional depends entirely on what happens next.

-- Kenji Nakamura, Tokyo