Short version: July's AI-agent funding did two opposite things at once, and the split is the whole point. By deal count, the money left Silicon Valley — funding trackers put roughly 42% of July's agent rounds outside it, with Paris, London, and Tel Aviv now behaving like real ecosystems. By dollar, it didn't move at all: the US still absorbed roughly 88% of H1 2026's AI venture dollars. That's not a contradiction — a market concentrates by dollar (a few enormous rounds) and disperses by deal (many small ones) in the same quarter. For a founder outside the Valley it reads as an instruction: build where you are, raise with a US door open.
The split: a wide map of deals, a narrow map of dollars#
Two numbers from July, and they point in opposite directions.
The deal map widened. As reported by funding trackers, about 42% of July's AI-agent deals closed outside Silicon Valley, and European weekly round-ups filled with agent-labeled Series As. The dollar map didn't. Roughly 88% of H1 2026 AI venture dollars still went to US companies — consistent with the concentration we covered when the US took 86 cents of every venture dollar.
Both are true because a dollar total is dominated by a handful of mega-rounds — one $5B raise outweighs five hundred $10M ones. So the number of places you can raise genuinely widened, while the largest checks stayed put. Read only the first number and you'll declare the Valley over; read only the second and you'll miss that a real ecosystem now exists where you live.
Where the deals are actually landing#
The dispersion is concentrated exactly where it's cheap to prove a company: seed and Series A. The named non-US rounds around July make the point better than the aggregate:
- Legora — a Stockholm legal-AI agent startup — raised a reported ~$550M Series D at a ~$5.55B valuation. It's an outlier by size, and that's the tell: even the big non-US round is a legal-vertical agent, echoing the control-the-agents-or-own-a-regulated-vertical thesis that defined the month's money.
- Tsuga — Paris, AI-native observability — landed a €30M Series A (EU-Startups).
- LinqAlpha — a ~$22M Series A among the early-July agent rounds.
And the hub story is real, not press-release geography. Paris has drawn roughly $5.8B in AI equity across ~187 rounds (Mistral is about half of it), enough for TechCrunch to call it the most important AI city outside Silicon Valley. London and Tel Aviv anchor their own follow-on chains; Stockholm, Berlin, Amsterdam, and Zurich fill in a credible European tier.
What it means: staying local is now a strategy, not a compromise. The density of capital, talent, and follow-on that used to exist in one place exists in several. If you're pre-Series-B, the ecosystem you need is probably reachable from where you already are.
Why the dollars still gravitate to one place#
The concentration isn't sentiment — it's structure. The biggest funds write the biggest checks, and the biggest funds still sit closest to the US market: deeper late-stage pools, the exit venues, and the customer base that underwrites a $280M-median markup (July's reported median agent post-money, up ~40% from Q1). A mega-round is a bet on a mega-outcome, and the machinery for that outcome — the acquirers, the IPO window, the growth capital — remains US-weighted.
What it means: the gravity is real but it's a cap-table problem, not a zip-code problem. You don't need to move your team to be legible to a US fund. You need to be structured so a US fund can lead your Series B without a painful re-incorporation.
If you're not in the Valley: build here, raise there#
The two-variable read for a non-US agent founder:
- Build local. Talent is cheaper outside the Valley, the local ecosystem is now thick enough to seed and Series-A into, and your early customers are probably closer to home. The deal map says you can.
- Raise with a US door open. Set up so a US lead is a decision, not a re-founding — many non-US teams use a US holding structure their target investors already understand, and build those investor relationships a year before the round. The dollar map says you'll want that door when the big round comes.
- Pick a vertical the mega-check respects. The largest non-US round of the month was a regulated-vertical agent. If you want to compete for the concentrated dollars from outside the concentrated geography, own a mess that's worth $5B to clean up — the same lesson as the $25M ARR line: July's money stopped paying for demos.
The old assumption was that serious money lived in one place, so serious founders moved there. July's data retires half of that: the serious deals are now in many places. The serious dollars still aren't — but that's the one part of the map you can navigate with a lawyer instead of a moving truck.
