In Q4 2025, AI and machine learning captured 52% of all global venture capital deal value — the first quarter on record where a single technology category claimed more than half. Total deal value that quarter was $138.6 billion, making AI's cut roughly $72 billion. Visual Capitalist mapped the full shift in their piece Where Venture Capital Money Is Going: AI vs. Everything Else.

How fast this happened

In 2020, AI's share of U.S. venture capital was 14%. By 2024 it had reached 45%. In Q1 2025 it hit 71% in the U.S. market alone. That is a five-year run from a minor allocation to a majority share.

To put it another way: biotech, fintech, clean energy, consumer tech, and every other sector combined now compete for less than half the available capital. That compression is structural, not a one-quarter blip.

Why capital moved this way

Venture investors follow returns. There are three concrete reasons AI attracted capital at this scale:

• Model training and inference infrastructure created measurable revenue for semiconductor and cloud companies. The thesis was grounded in actual earnings, not just speculation.
• Enterprise software companies found that adding AI features justified price increases. That changed the margin story for a category that had been commoditizing.
• Generative AI tools compressed development timelines for small teams, making early-stage investments faster to de-risk than in previous cycles.

What 52% looks like from the other side

For sectors outside AI, this concentration means capital is genuinely scarcer. Founders in biotech or clean energy who cannot frame their story around AI integration are raising in a thinner market than they were three years ago. That is not a sentiment shift — it shows up in deal counts and round sizes.

For public market investors watching from a distance, the implications are different. Smart money moving at this scale into a category tends to move prices before the conviction becomes widely shared. Buying AI exposure today means buying into expectations that VC has already been pricing in for 18 months.

Is 52% a ceiling or a floor?

The U.S. figure hit 71% in Q1 2025. Historical data on sector concentration in VC suggests that once a single category reaches that level, mean reversion tends to follow — either because valuations get stretched, or because adjacent sectors attract investors looking for relative value. But "tends to follow" is not a timing signal. Infrastructure demand has been backing the AI thesis with real revenue, and that distinction matters when comparing it to cycles that were driven by narrative alone.

The 52% figure does not tell you whether to buy or avoid AI. It does tell you that the trade is no longer early. That is worth knowing.