The useful read
hub vs hub
┌─ generated from funding, org, model, compute, and overlap signals ─────────────────────────────── ┐
New York and Paris are close on tracked AI funding, so the interesting split is models, compute, and org depth.
New York's biggest carrier is Reflection: $2.1B (30.2% of tracked side funding). Paris's biggest carrier is Mistral AI: $3.0B (42.4% of tracked side funding).
The org treemap is shown because both sides have enough funded orgs for area to mean something.
Funding is HQ-based and limited to AI-native orgs. Compute is counted by datacenter location, so it can disagree with headquarters funding.