AI Venture Funding Takes Two-Thirds as North America Drops 35%

What happened

North American startups raised $92 billion across seed through growth-stage rounds in the third quarter of 2026, according to Crunchbase data covering U.S. and Canadian companies. 2 That is a 35% drop from the previous quarter but a 50% increase over the same period a year earlier. 12 On its face, a one-third sequential decline looks like a sharp pullback. The underlying data shows something different: capital has become even more concentrated in artificial intelligence.

Roughly two-thirds of the quarter's North American venture dollars went to AI-focused companies, about $61 billion by one estimate. 12 Crunchbase attributes most of the quarter-over-quarter decline to a single factor. The quarter had no fresh megarounds for OpenAI and Anthropic, and the firm says the dip does not signal a broad weakening in venture investment. 2 Deal volume supports that view. Counts of both early- and late-stage rounds stayed close to the prior quarter's levels. 2

The global picture: fewer dollars, bigger checks

The pattern holds worldwide. Global venture funding fell 25% from the $212 billion raised in Q2 but rose 53% from $104 billion in Q3 2025. 4 U.S. companies took $91 billion, about 57% of the global total, and the San Francisco Bay Area alone accounted for 24%. 3 AI startups across the stack raised $102 billion globally, about 64% of all venture capital. 3

The quarter also set a record for the number of billion-dollar rounds. 4 Databricks and Safe Superintelligence each raised $5 billion. Crusoe, Moonshot AI, Mistral AI, Nscale, The Boring Co. and Kling AI each raised more than $3 billion. 4 Databricks, Safe Superintelligence and Crusoe made up a large share of North America's AI total. 2 By one count, a third of the quarter's capital went to just 27 companies, leaving the remaining two-thirds spread across thousands of rounds. 3 Late-stage funding reached $105 billion globally, down 23% from Q2 but up 73% from a year earlier. 4

Where the accounts diverge: early stage

The sources describe early-stage activity differently. One summary says early-stage funding stayed well above year-ago levels even after slipping from Q2. It reads this as evidence that seed and Series A investors have not pulled back, despite fatigue around late-stage megarounds. 1 Crunchbase's own account stresses that early-stage dealmaking fell sharply from a multiyear peak the quarter before. 2

Both statements can be true. Q2 was an unusually high benchmark, and a steep drop from a peak can still leave totals above last year's. The more telling detail is at seed. Global seed funding totaled $13 billion, and $2.6 billion of that, around a fifth, came from seed rounds of $100 million or more. 4 Some "seed" rounds are now as large as growth rounds, so stage labels say less about company maturity than they once did.

The squeeze on everyone else

For founders outside AI, the averages are misleading. Carta data shows a foundational-model AI startup at Series A might raise at a median valuation of $300 million. A non-AI startup at the same stage sits around $55 million. 3 Half of the quarter's capital went to companies founded since 2022, a sign of how fast investors are backing new AI-native teams. 3 One analysis put it this way: non-AI startups are competing for a shrinking slice of a shrinking pie. 1

Exits remain the weak point. Only 17 venture-backed North American companies went public in Q3, raising less than $4 billion combined, and none of the listings was a blockbuster. 1 Crunchbase notes that the largest AI companies are eyeing the public markets. 2 So far, however, the capital going into these companies far outpaces the liquidity coming out.

The reading

The 35% drop should not be read as a cooling of venture appetite. Overall activity is still well above last year, deal counts are stable, and billion-dollar rounds hit a record. The quarterly totals mostly track how often a few frontier labs and AI infrastructure firms raise money. When OpenAI or Anthropic closes a round, the numbers jump. When neither does, they fall.

The structural story is concentration in a few large, mostly Bay Area-based AI companies, with valuations that leave non-AI startups far behind. That makes the market fragile in a specific way. The headline numbers depend on a handful of companies and their investors, and the exit route that would return cash to those investors is still mostly hypothetical. If the AI giants do go public and are well received, the cycle can keep going. If not, today's extreme concentration becomes a concentrated risk.