Falcoscan Atlas maps 0 AI tools across 19 active global hubs (of 19 mapped); 2,350 carry full audit-trail enrichment. San Francisco Bay Area leads with 1,293 tools, dominant in Agents — roughly 60% of the geo-attributed sample. NYC, London, Boston, Tel Aviv round out the top five. The parent catalog tracks 6,702 tools across 29 markets; this view zooms in on where the most-watched are headquartered.
20 hubs. 0 AI tools mapped.
Drawn from Falcoscan’s 6,702-tool catalog across 29 markets. Atlas surfaces only tools with verified HQ data. Coverage expands quarterly.
MethodologyTools mapped: every tool with a verified HQ. 2,350 carry full audit-trail enrichment (confidence ≥ 50). Coverage: 19 of 20 hubs · 0 tools placed. Refreshed .
| Rank | Hub | Country | Tracked tools | Dominant category |
|---|---|---|---|---|
| 1 | San Francisco Bay Area | United States | 1293 | Agents |
| 2 | New York City | United States | 170 | Ecommerce |
| 3 | Greater London | United Kingdom | 133 | Image |
| 4 | Greater Boston | United States | 98 | Marketing |
| 5 | Tel Aviv-Yafo | Israel | 84 | Security |
| 6 | Seattle Metro | United States | 79 | AI Models |
| 7 | Île-de-France | France | 71 | AI Models |
| 8 | Greater Toronto | Canada | 63 | AI Models |
| 9 | Bangalore | India | 35 | Voice |
| 10 | Berlin | Germany | 29 | Agents |
| 11 | Amsterdam | Netherlands | 24 | Design |
| 12 | Los Angeles | United States | 23 | Ecommerce |
| 13 | Greater Tokyo | Japan | 15 | Voice |
| 14 | Austin | United States | 15 | Writing |
| 15 | Beijing | China | 13 | Video |
| 16 | Stockholm | Sweden | 9 | Legal |
| 17 | Singapore | Singapore | 7 | Design |
| 18 | Shenzhen | China | 6 | AI Models |
| 19 | Zürich | Switzerland | 4 | Design |
Bay Area concentrates an estimated $50.4B in tracked AI capital across 1,901 funded tools — 56% of the $90.3B aggregated across 3,439 funded tools we track. NYC, the next-largest, holds $8.3B (9%). The capital concentration of AI is more extreme than the tool concentration above.
The chart above tells the headline. Every hub in the index, with heat band, dominant category, and the two scores that ground each deep-dive — opportunity (where to build) and saturation (where the field is already crowded).
Foundation models, infrastructure, frontier labs
“Where the AI economy is denominated. Saturated at the top, brutal in the middle.”
Applied AI, enterprise, vertical SaaS
“NYC sells the picks and shovels. Less glamour, better margins.”
Research labs, video AI, fintech-native AI
“London is where research becomes product, slowly and well.”
Open source, foundation models, sovereign AI
“Paris owns open source. Everyone else is renting.”
Cybersecurity AI, dev tools, agentic infra
“Tel Aviv ships infrastructure. The unsexy parts everyone needs.”
Deep learning research, healthcare AI
“Toronto trained the founders of half the AI economy.”
Bio AI, robotics, academic spinouts
“Boston is where AI meets atoms.”
Cloud AI, voice, large-org tooling
“Seattle is the quiet AI capital. Built into the pipes.”
Open source, dev tools, enterprise AI
“Berlin builds for builders. Practical, slow-burning, durable.”
APAC distribution, fintech AI, regulated AI
“Singapore is the regulated-AI on-ramp into Asia.”
Applied AI, services-to-product, BPO disruption
“Bangalore rewrites the services economy in code.”
Frontier models, computer vision, autonomy
“Beijing builds for the largest market in the world. Different rules apply.”
Hardware AI, robotics, edge inference
“Shenzhen is where AI meets manufacturing speed.”
Robotics, industrial AI, anime/creative
“Tokyo ships robots. Everyone else ships pitch decks.”
Open source, audio AI, fintech AI
“Stockholm punches three weight classes above its size.”
Search, recommendation, payments AI
“Amsterdam is where European AI scales commercially.”
Defense AI, robotics, capital-light infra
“Austin took the Bay Area exodus and built something different.”
Creative AI, video, agentic media
“LA is where AI meets entertainment. Volume over rigor.”
Hubs with fewer than five tracked tools — present in the map and the Sankey, no full editorial treatment until coverage expands: Zurich(4).
Fully distributed teams with no fixed HQ
Foundation models versus open source. Enterprise pricing versus prosumer freemium. The same dataset, two perspectives.
Each hub on the left flows into the categories its tools cluster in. Ribbon weight equals tool count. Edges below three tools and hubs producing fewer than three distinct edges are excluded for legibility.
Sankey diagram showing each AI hub on the left flowing into the categories its tools cluster in. 10 hubs, 13 categories, 49 flow edges.
| Hub | Category | Tools |
|---|---|---|
| bangalore | Writing | 6 |
| bangalore | Support | 4 |
| bay-area | Other (15) | 146 |
| bay-area | AI Models | 59 |
| bay-area | Agents | 55 |
| bay-area | Coding | 46 |
| bay-area | Productivity | 45 |
| bay-area | Image | 42 |
| bay-area | Finance | 39 |
| bay-area | Video | 30 |
| bay-area | Data | 30 |
| bay-area | Support | 28 |
| bay-area | Sales | 23 |
| bay-area | Voice | 21 |
| bay-area | Writing | 18 |
| berlin | Writing | 6 |
| berlin | Productivity | 3 |
| boston | Other (15) | 24 |
| boston | Support | 4 |
| boston | Writing | 3 |
| london | Other (15) | 15 |
| london | Voice | 11 |
| london | Image | 7 |
| london | Video | 3 |
| new-york | Other (15) | 22 |
| new-york | Voice | 9 |
| new-york | Finance | 7 |
| new-york | Video | 6 |
| new-york | Agents | 5 |
| new-york | Data | 4 |
| new-york | Support | 4 |
| new-york | AI Models | 3 |
| new-york | Sales | 3 |
| new-york | Coding | 3 |
| paris | Other (15) | 8 |
| paris | Image | 4 |
| paris | AI Models | 3 |
| paris | Productivity | 3 |
| seattle | Sales | 7 |
| seattle | AI Models | 5 |
| seattle | Agents | 5 |
| seattle | Coding | 3 |
| tel-aviv | Other (15) | 10 |
| tel-aviv | Video | 5 |
| tel-aviv | Image | 3 |
| toronto | Image | 7 |
| toronto | AI Models | 6 |
| toronto | Video | 4 |
| toronto | Other (15) | 4 |
Hubs producing fewer than three distinct categorized edges, and edges below three tools, are excluded for clarity. Coverage expands with each quarterly enrichment cycle.
The Q2 2026 edition publishes once the current enrichment cycle completes. Last underlying-data refresh: pending.
Falcoscan Editorial · Updated quarterly
Get the next Atlas in your inbox.