The New York AI Map: 22 Companies Building the Application Layer

The New York AI Map

Ramp is worth $44B. VAST Data $30B. ElevenLabs $11B.

None of them are headquartered in San Francisco.

New York has quietly assembled an AI cluster with a completely different shape to the Bay Area's, and the difference tells you something useful about where value is accruing in this cycle.

The 22 Companies

Frontier and infrastructure. Reflection AI, a frontier lab from ex-DeepMind researchers that raised $2B led by Nvidia. Hugging Face, the home of open-source AI. Modal at $4.65B. Pinecone in vector databases. VAST Data at $30B.

Finance AI. Ramp at $44B in finance operations. Hebbia in document intelligence. Rogo in investment banking, backed by Sequoia. AlphaSense at $7.5B in market intelligence. Basis at $1.15B in accounting agents.

Media and voice. Runway at $5.3B in AI video. ElevenLabs at $11B in voice. Captions with more than 20 million users in video editing.

Health and compliance. K Health in primary care. Tennr in healthcare workflow. Norm Ai at $1.2B in compliance.

Go-to-market and sales. Clay at $3.1B. Attention in revenue operations. Regal in voice agents.

Enterprise AI. UiPath at $1.85B ARR in agentic automation. EliseAI at $2.2B. Dataiku with $350M+ ARR.

Three Things That Stand Out

New York wins the application layer

San Francisco builds the models. New York builds the things that make them pay: finance, law, healthcare, sales.

That split is not an accident of geography. It follows the buyers. If you are building AI for investment banking, being in the same city as the investment banks is worth more than being near the researchers. The feedback loop that matters runs through customers, not through papers.

The money is real

Over 1,000 NYC AI companies have raised more than $27 billion since 2019.

Worth noting the quality of the revenue, not just the valuations. UiPath reports $1.85B in ARR. Dataiku reports over $350M. In a market where most AI valuations rest on projections, New York's cluster has an unusual concentration of companies with disclosed, substantial, recurring revenue.

Flatiron is becoming what SoMa was in 2012

Clay, Ramp, Basis, Dataiku and Captions all sit within ten blocks of Madison Square Park.

Density like that produces the thing that is genuinely hard to manufacture: people leaving one company to start the next one, with the network intact.

Why This Matters If You Run Go-To-Market

The temptation with a map like this is to read it as tech news. There is a more practical reading.

Every company on the right-hand side of that split is a building block you can wire into how you find, reach, and close customers. Clay for data orchestration. Attention for revenue operations. Regal for voice. ElevenLabs if you are building anything with audio.

The teams that win will not be the ones with the most tools. They will be the ones who know which block does what, and who wire the blocks together rather than buying another one.

That is the actual lesson from the application layer. Every company in it succeeded by owning a workflow completely rather than adding a feature to someone else's. The same logic applies to your own stack. A tool that improves one step leaves you holding the integration. A system that owns the whole job does not.

The Broader Pattern

The model layer consolidated into a duopoly because scale wins there. The application layer fragmented into dozens of defensible niches because domain knowledge wins there.

For most businesses, only the second one is a place you can compete. You are not going to out-train Anthropic. You can absolutely know more about a specific workflow than a general model does, and turn that into a product.

New York's cluster is what it looks like when a city bets entirely on that second thing.

New York AI By Sector

Frequently Asked Questions

Why is finance the largest cluster in New York?

Proximity to buyers. Building AI for investment banking is faster when you can have coffee with investment bankers, watch the workflow, and get corrected quickly. That feedback loop matters more in applied AI than proximity to researchers does.

Is New York competing with San Francisco?

Not directly, and that is the point. San Francisco builds the models. New York builds the applications that make them pay. They are complementary layers, and the New York companies mostly rent their models from the San Francisco ones.

Which of these matter for a go-to-market team?

Clay most directly, as the data orchestration layer in a lot of modern outbound stacks. ElevenLabs if you are building anything with voice. Attention and Regal in revenue operations and voice agents respectively. Hugging Face if you run open models.

How much money is actually in the NYC AI scene?

More than 1,000 NYC AI companies have raised over $27 billion since 2019. Unusually for this market, several of the largest have substantial disclosed recurring revenue rather than only valuations: UiPath reports $1.85B ARR and Dataiku over $350M.

What makes the Flatiron concentration significant?

Clay, Ramp, Basis, Dataiku and Captions all sit within ten blocks of Madison Square Park. That density produces the thing that is hard to manufacture deliberately: people leaving one company to start the next one with their network intact.

We build AI-native pipeline systems and you pay per qualified meeting, not a retainer. No paying for activity. You only pay when we book you a qualified sales meeting. See if you qualify.

Valuations and funding figures reflect publicly reported data as of mid-2026. ARR figures are company-stated. Not investment advice.