New York City is no longer simply a place where companies use artificial intelligence. It has become one of the most important places in the world to build AI companies.
That change has happened quickly. A few years ago, the usual story was that San Francisco built the AI models while New York applied them to banking, advertising, healthcare, media and other industries. That picture is now too simple. New York still has a major edge in applied AI, but the city is also producing frontier research labs, AI infrastructure companies, world-model developers and companies trying to rethink how artificial intelligence itself works.
The numbers show the size of the shift. New York City has more than 2,000 AI startups, more than 40,000 workers with AI skills, over 25,000 technology startups overall, and more than 1,200 active venture capital firms, according to city data. NYC universities including Columbia, Cornell Tech, CUNY and NYU produced more than 87,000 graduates with AI-ready degrees between 2018 and 2023.
The money is moving in the same direction. Tech:NYC reported that NYC-based AI companies raised $15.84 billion in 2025, up 50% from the previous year. AI companies also leased more than 486,000 square feet of Manhattan office space during the year, an important sign that the growth is creating a physical business cluster rather than only remote startups with New York mailing addresses.
The momentum continued into 2026. New York startups raised $8.88 billion in Q2 2026, according to AlleyWatch. In June alone, the city captured 24.4% of all U.S. venture dollars, and large rounds went to several AI companies covered in this article, including Flourish, AlphaSense, General Intuition and Rogo.
There is another reason businesses should pay attention. AI-related job postings in New York increased 120% year over year, while AI adoption among executives surveyed by Tech:NYC and Accenture rose from 55% to 78%. Almost every executive in the survey—99%—said their organization planned to increase recruiting for AI-related roles.
This means the most interesting question is no longer, “Does New York have an AI industry?”
It clearly does.
The better question for 2026 is: Which New York AI companies are building something that can become much larger?
We researched dozens of candidates and narrowed the market to 20 companies worth watching closely. The result includes established unicorns, fast-growing business software companies, young research labs, infrastructure providers and highly focused startups attacking expensive problems in industries where New York already has an advantage.
The 20 Top AI Companies in NYC to Watch in 2026
This is not simply a ranking of companies by valuation. A company with a huge funding round but no customers should not automatically rank above a smaller business solving an expensive problem for hundreds of paying companies.
Instead, we looked for four things: strong New York roots, AI that is central to the product rather than a small feature, evidence of real technical or commercial progress, and a reason the company could become more important during the next several years.
| Company | Main AI Market | Recent Public Signal | Why It Matters |
|---|---|---|---|
| Reflection AI | Frontier/open AI | Close to $2.6B reported funding | New York now has a serious frontier-model contender. |
| AlphaSense | Market intelligence | $350M raised at $7.5B valuation; $600M+ ARR | One of NYC’s clearest examples of AI becoming large enterprise software. |
| Runway | AI video/world models | $315M Series E at $5.3B valuation | Moving from creative AI toward broader world models. |
| Hugging Face | Open AI infrastructure | $395.2M raised; acquisition interest reported | One of the world’s most important open AI platforms is headquartered in Brooklyn. |
| General Intuition | Physical AI/world models | $320M round at $2.3B valuation | Uses gameplay data to teach AI how to act in physical environments. |
| Flourish | Neuro-AI research | $500M financing; reported $2.5B valuation | A major attempt to build more efficient AI by studying how brains learn. |
| EliseAI | Housing and healthcare | $250M Series E; $100M+ ARR reported in 2025 | Strong proof that vertical AI can automate complex real-world operations. |
| Rogo | Financial AI | $160M Series D; $300M+ total funding | Built specifically for investment banks and other financial institutions. |
| Norm AI | Legal and compliance AI | $120M Series C at $1.2B valuation | Building AI agents around law, regulation and compliance. |
| OpenRouter | AI model infrastructure | $113M Series B; joining Stripe announced Aug. 19 | Its exit shows how valuable the model-routing layer has become. |
| Rillet | AI accounting/ERP | $100M Series C at $1B valuation | Wants to replace old accounting systems with an AI-native finance platform. |
| Mirage | AI video | $75M growth financing; $175M+ total funding | Building AI video tools used by creators and businesses. |
| Hebbia | Enterprise research AI | $130M Series B; Matrix 2.0 launched in 2026 | Competing to become the AI workbench for document-heavy knowledge work. |
| Tennr | Healthcare workflow AI | $101M Series C | Uses purpose-built models to fix specialist referral workflows. |
| Pinecone | Vector database/AI infrastructure | $138M total funding; 5,000+ customers | Infrastructure used to make AI applications more useful with private data. |
| Viam | Physical AI and automation | $30M Series C; $117M total funding | Connects software, data, machines and AI in the physical world. |
| Nayya | Benefits and health AI | $130M funding; nearly 1M users reported | Shows how AI can simplify complicated employee benefits decisions. |
| Reserv | Insurance claims AI | $125M Series C | Rebuilding claims operations around AI rather than adding AI to old systems. |
| TollBit | AI content licensing | $24M Series A; about $31M total disclosed funding | Building infrastructure for publishers to charge AI systems for content access. |
| Arthur AI | AI evaluation and governance | $60M+ disclosed funding | Focused on the growing problem of testing, monitoring and controlling AI systems. |
Original Research: What Our 20-Company NYC AI Dataset Tells Us
Lists of AI startups are easy to create. Useful analysis is harder.
For this article, NYC Tech Journal created a dataset covering the 20 companies above and compared their funding, market category, financing dates, business focus and public signs of commercial adoption. The goal was not to create a fake scientific ranking from private-company data that is often incomplete. The goal was to find patterns that tell founders, business leaders, workers and investors what is actually happening inside New York’s AI economy.
Our Research Method
The research cutoff was August 28, 2026. We gave priority to company announcements, New York City government and economic-development data, and direct funding announcements, followed by reporting from publications such as TechCrunch, Forbes, WIRED, Fortune, Axios and Business Insider.
Private-company numbers require care. Funding databases can disagree, valuations can change between rounds, and companies sometimes describe funding as “more than” a certain amount. For our combined funding calculations, we used the minimum amount we could reasonably verify when a source said “more than” or “over.” We excluded financing rounds that were only reported as being discussed and did not count acquisition prices as venture funding.
That makes our combined funding estimate intentionally conservative.
Finding #1: These 20 Companies Have Raised Roughly $8.3 Billion
Using the method above, the 20 companies in our sample account for approximately $8.3 billion in publicly disclosed or credibly reported funding.
That figure should be read as a lower-bound estimate rather than a perfect accounting total. AlphaSense, for example, says its total funding is “well over $1 billion,” but we counted $1 billion for this calculation. Reflection AI was reported in July 2026 to have raised close to $2.6 billion, so we used approximately $2.6 billion rather than trying to estimate undisclosed dollars.
Chart: Approximate Funding by AI Segment in Our NYC Sample
| Segment | Companies | Share of Companies | Approx. Funding | Share of Sample Funding |
|---|---|---|---|---|
| Core AI models and infrastructure | 6 | 30% | $3.86B | 46.6% |
| Finance, legal and enterprise knowledge | 5 | 25% | $1.93B | 23.3% |
| Creative media and web economics | 3 | 15% | $1.07B | 12.9% |
| Health, housing, benefits and insurance | 4 | 20% | $862M | 10.4% |
| Physical AI and robotics | 2 | 10% | $571M | 6.9% |
| Total | 20 | 100% | ≈$8.3B | 100% |
There is an interesting tension inside these numbers. Only 30% of the companies in our sample sit in what we classify as core models or AI infrastructure, but they account for almost 47% of the funding.
That is partly the Reflection AI effect. Frontier AI requires enormous amounts of money for computing, data, researchers and infrastructure. Reflection has already signed extremely large compute commitments, including a reported SpaceX arrangement worth as much as $6.3 billion over its potential term and a separate $1 billion Nebius compute deal.
The broader lesson is more useful than the raw funding number: New York is producing many applied-AI companies, but frontier AI is starting to attract frontier-sized capital in the city too.
Finding #2: 70% of Our Companies Are Applied or Physical AI Businesses
Six companies in our sample primarily sit in the core AI model or infrastructure layer. The other 14, or 70%, use AI to solve a defined business or physical-world problem.
That fits New York unusually well.
NYC does not need to beat Silicon Valley at every part of AI to become one of the most important AI markets. It can win by bringing artificial intelligence directly into fields where New York already has deep customers, talent and specialized knowledge: banking, private equity, accounting, law, insurance, media, healthcare, real estate and advertising.
NYCEDC has made the same broader bet. Its AI strategy explicitly focuses on making New York a global leader in applied AI, using the city’s industry diversity as an advantage rather than trying to copy another technology hub.
Our company sample gives that idea real form. Rogo sits close to investment banks. Norm AI can recruit legal and regulatory talent. EliseAI is surrounded by large landlords, healthcare systems and property companies. AlphaSense sells to many of the institutions that already make New York a global business center.
This is not location as branding. It can become location as product advantage.
Finding #3: Capital Is Highly Concentrated
Our five most heavily funded companies account for roughly $5.4 billion, or about 65% of all disclosed funding in the 20-company sample.
| Funding Group | Approx. Funding | Share |
|---|---|---|
| Top five companies | $5.41B | 65.3% |
| Remaining 15 companies | $2.87B | 34.7% |
| Total | $8.29B | 100% |
This matters because big AI funding totals can give a misleading picture of how easy fundraising is.
New York’s venture market is strong, but capital is not being spread evenly. AlleyWatch found a similar pattern across the wider NYC startup market in Q2 2026: late-stage deals represented only 18% of deals but absorbed 69% of capital.
For founders, the message is simple. Do not look at $300 million and $500 million AI rounds and assume investors have stopped caring about proof. The opposite may be true. When a small number of huge deals soak up a large share of available dollars, everyone else needs a clearer case for why customers will pay.
Finding #4: At Least $2.18 Billion Was Raised in Ten Clearly Verified 2026 Rounds
Ten companies in our sample announced financing rounds during 2026 that we could cleanly identify and compare. Those rounds alone total $2.178 billion.
The median round was $142.5 million, while the average was $217.8 million. The average being far higher than the median tells us that the biggest deals are pulling the total upward.
Chart: Major 2026 Funding Rounds in Our Sample
| Company | 2026 Financing | Relative Size |
|---|---|---|
| Flourish | $500M | ████████████████████ |
| AlphaSense | $350M | ██████████████ |
| General Intuition | $320M | █████████████ |
| Runway | $315M | █████████████ |
| Rogo | $160M | ██████ |
| Reserv | $125M | █████ |
| Norm AI | $120M | █████ |
| OpenRouter | $113M | ████ |
| Rillet | $100M | ████ |
| Mirage | $75M | ███ |
| Total | $2.178B |
These ten deals are particularly important because they cover very different parts of AI. The money is not going only into chatbots. It is going into neuroscience research, financial analysis, accounting, insurance claims, legal work, video creation, model infrastructure and machines that can understand physical environments.
That variety makes the New York ecosystem more interesting than a single funding number suggests.
1. Reflection AI
Reflection AI is one of the strongest signs that New York’s AI market is entering a new phase.
The Brooklyn company was founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou. It originally worked on autonomous coding agents but expanded its ambition toward building frontier-scale open AI systems. In October 2025, Reflection announced a $2 billion raise at an $8 billion valuation.
By July 2026, TechCrunch reported that Reflection had raised close to $2.6 billion and was being valued at roughly $25 billion on a pre-money basis. It was also committing huge sums to computing infrastructure. That is the kind of capital profile normally associated with a small group of West Coast frontier labs.
Why Reflection AI Matters for NYC
Reflection changes the story of what an AI company in New York can be.
The city’s strength has traditionally been turning technology into useful business products. Reflection is trying something different: building models and training infrastructure near the base of the AI stack itself.
That creates opportunities beyond Reflection. Frontier labs attract researchers, infrastructure companies, investors and specialized workers. They also make New York more credible to technical founders who previously assumed that building a model company meant moving to San Francisco.
The risk is equally large. Frontier models burn enormous amounts of capital before anyone knows how durable their advantage will be. Reflection therefore belongs near the top of this list because both the upside and the execution challenge are enormous.
2. AlphaSense
If Reflection represents New York moving deeper into frontier AI, AlphaSense represents the city’s existing strength in applied enterprise AI.
AlphaSense helps professionals search, understand and act on financial and business information. Its system brings together company filings, earnings calls, research, expert information and other sources, then uses AI to help users find what matters.
In June 2026, AlphaSense raised $350 million at a $7.5 billion valuation. More important than the valuation was the operating number released alongside it: the company said annual recurring revenue had passed $600 million in Q1 2026.
AlphaSense also says more than 7,000 enterprises use its platform, including 90% of the S&P 100 and all of the world’s top global investment banks.
Why AlphaSense Matters for NYC
Many AI startups still have to prove that businesses will pay large amounts of money for their products. AlphaSense has largely moved beyond that question.
Its challenge is now becoming part of the operating system for high-value business decisions.
That is why its June launch of SuperAnalyst matters. Rather than simply returning answers to research questions, the product is designed to execute multi-step projects, monitor developments and create decision-ready work.
For New York businesses, AlphaSense is also a useful model to study. The company did not start by trying to automate every job. It built around one expensive problem—finding reliable market intelligence—and kept expanding the workflow around it.
3. Runway
Runway began as one of the best-known names in generative video. In 2026, it is trying to become something broader.
The New York company raised a $315 million Series E in February 2026, with the round valuing it at about $5.3 billion. Crunchbase News estimates Runway has raised approximately $860 million since it was founded in 2018.
Its next target is world models. These systems try to understand how environments behave rather than simply predict the next word in a block of text.
That could make Runway relevant far outside filmmaking. The company has discussed uses stretching into gaming, robotics, medicine, climate and energy, while continuing to develop its creative-video business.
Why Runway Matters for NYC
Runway is one of the clearest examples of New York’s creative industries and AI research reinforcing each other.
The city contains large advertising agencies, media companies, brands, filmmakers, production businesses and design teams. Those organizations create a natural testing ground for generative video.
But the bigger 2026 question is whether Runway can use the knowledge it gained from video generation to build more general systems that understand how the world works.
If it succeeds, it could move from being seen mainly as an AI video company to becoming a much broader AI research company. That is a far larger opportunity, but it also puts Runway into competition with some of the best-funded laboratories on earth.
4. Hugging Face
Hugging Face is one of the most important AI companies that many non-technical business leaders have barely heard about.
Headquartered in Brooklyn, Hugging Face operates a platform where developers and researchers share AI models, datasets, applications and tools. Forbes reported in 2025 that the company had roughly 10 million users, content from hundreds of thousands of organizations and $395.2 million in total funding.
That makes Hugging Face less like a single AI application and more like infrastructure for an enormous community.
A Major M&A Story Is Developing
Hugging Face is also one of the fastest-moving names in this article.
As of our August 28, 2026 research cutoff, multiple outlets have reported acquisition interest, including reports linking Nvidia to a possible transaction around $13 billion. However, reporting on whether a final agreement exists has been inconsistent, and neither company had publicly confirmed a completed acquisition in the sources we reviewed. Business Insider reported that no deal had been finalized.
We therefore treat Hugging Face as an independent company for this analysis while clearly flagging that its ownership could change.
Why Hugging Face Matters for NYC
Hugging Face gives New York something every technology ecosystem wants: a product that developers around the world already use.
That creates network effects that are difficult to reproduce. The more researchers publish models and datasets there, the more useful the platform becomes to developers, which attracts still more contributors.
Its expansion into robotics is also worth watching. Hugging Face acquired Pollen Robotics and moved into hardware with Reachy Mini, showing that its ambitions now extend beyond hosting software models.
5. General Intuition
General Intuition may be one of the strangest—and most interesting—AI bets being made in New York.
The company was spun out of Medal, a platform for sharing video-game clips. That history gave General Intuition something most robotics labs do not have: enormous amounts of gameplay data paired with the actual button presses that caused each action.
The company believes those action labels can help AI understand how to move through space and time.
General Intuition raised $320 million at a $2.3 billion valuation in June 2026, bringing disclosed funding to $454 million. During a TechCrunch visit to its New York office, the company demonstrated a model running both a game agent and a physical robot.
Why General Intuition Is One to Watch Closely
On August 24, TechCrunch reported that General Intuition was already discussing another financing at a $6 billion pre-money valuation. Because that financing was still being finalized, we did not include it in our funding calculations.
The speed is remarkable. General Intuition only spun out in late 2025.
The bigger reason to pay attention is its thesis. Most generative AI became powerful by learning from huge amounts of human language. General Intuition believes machines that need to act in the physical world require a different kind of training data.
If that idea proves correct, gaming data could become far more valuable than it first appears.
6. Flourish
Flourish is one of the boldest AI research companies to appear in New York during 2026.
The company is studying the brain in search of ideas that could make artificial intelligence more energy efficient and better at learning continuously. WIRED reported that Flourish has $500 million in funding and a reported $2.5 billion valuation, with backers including Jeff Bezos, Lux Capital and GV.
The company is operating from a large SoHo research facility. AlleyWatch described the operation as a ten-story lab, making Flourish a useful example of something important happening in NYC: serious AI research is starting to occupy physical laboratory space in Manhattan.
Why Flourish Is Different
Most current AI development starts with the architecture already proven by modern machine learning and tries to make it bigger, faster or cheaper.
Flourish is asking whether researchers should borrow more directly from biology.
The thesis sounds simple. The human brain learns continuously while using a tiny fraction of the energy consumed by large AI computing systems. If researchers can understand useful parts of the brain’s learning process and recreate them in software, AI systems could potentially become far more efficient.
That is a huge “if.” Flourish is still a research bet rather than a mature software company with a proven commercial machine.
But that is exactly why it belongs on a watch list.
7. EliseAI
EliseAI shows what happens when an AI startup picks painful industries instead of easy demos.
The New York company develops AI systems for housing and healthcare operations. Its tools can handle conversations, scheduling, leasing work, collections and other administrative processes where businesses often depend on large teams doing repetitive tasks.
EliseAI raised $250 million in Series E funding in August 2025. At the time, it said it had passed $100 million in annual recurring revenue, worked with more than 600 housing owners and operators, and served 75% of the NMHC Top 50 operators.
The company has continued growing. Its current website says one in six U.S. apartments uses EliseAI, that it supports two million patients, has raised more than $400 million and has reached $200 million in ARR.
Why EliseAI Is Strategically Important
Housing and healthcare share several features that make them good markets for vertical AI.
Both are complicated. Both involve huge amounts of communication. Both have old software. Both have rules that a generic chatbot cannot safely ignore.
EliseAI’s growth suggests that the most valuable AI agent companies may not look like general assistants. They may look like highly trained digital workers built around one industry’s exact processes.
That lesson is particularly important in New York, where specialized industries exist at enormous scale.
8. Rogo
Wall Street creates a natural laboratory for AI.
Analysts, bankers and investors spend enormous amounts of time reading filings, searching company information, building models, preparing presentations and reviewing documents. Those tasks are expensive because the workers doing them are expensive.
Rogo is building specifically for that environment.
The New York company raised $160 million in Series D funding in April 2026, bringing total funding to more than $300 million. It says its platform is used across more than 250 investment banks and investment firms.
Its agent, Felix, is designed to perform multi-step finance work rather than simply answer isolated questions.
Why Rogo Has a Real NYC Advantage
Rogo is a textbook example of why applied AI could cluster in New York.
Its customers are not far away. Some are within walking distance.
That matters more than it may seem. Building serious enterprise AI requires understanding how work actually happens, where data lives, which mistakes are unacceptable and what users will trust an agent to do.
Financial institutions are also unusually demanding buyers. A product that survives their requirements for security, accuracy and control can become stronger because of those requirements.
Rogo still faces serious competition from firms such as Hebbia and broader enterprise AI platforms. But finance is big enough for several major businesses to emerge.
9. Norm AI
AI creates a strange problem for companies.
Businesses want agents that can act with more independence. At the same time, laws and regulations place limits on what those systems should be allowed to do.
Norm AI wants to build the bridge between those two forces.
The New York company raised a $120 million Series C at a $1.2 billion valuation in July 2026, bringing total funding to more than $260 million. Norm describes its approach as “agentic law,” which means embedding legal and compliance rules directly into AI-driven workflows.
Its broader structure is unusual. Alongside its technology business, Norm has an affiliated AI-native law firm, Norm Law, where attorneys supervise and improve systems being used for legal work.
Why Norm AI Could Become More Important as Agents Grow
The first wave of business AI focused heavily on generating content.
The next wave involves software taking actions.
That makes compliance far more important. A chatbot drafting an internal summary creates one level of risk. An autonomous system making decisions, communicating externally or carrying out regulated workflows creates a very different one.
Norm’s timing therefore makes sense. The more powerful agents become, the more companies need ways to limit them.
New York’s concentration of banks, insurers, law firms and regulated companies gives Norm a rich local customer market.
10. OpenRouter
OpenRouter provides a different type of AI infrastructure.
Instead of forcing developers to build around one model company, its platform makes it easier to access, compare and route requests across many AI models through a common system.
That position became extremely valuable as the number of strong models exploded.
OpenRouter announced a $113 million Series B in May 2026. Then, on August 19, the company announced that it was joining Stripe. Axios reported that the transaction was worth more than $8 billion in cash and stock.
Why We Still Included OpenRouter
Strictly speaking, OpenRouter is now a different type of entry from the independent startups elsewhere in this article.
We kept it because the deal is one of the most important pieces of evidence about New York’s AI ecosystem in 2026.
A company founded only a few years earlier built a valuable position between developers and model providers. Stripe appears to see that routing layer as strategically important enough to spend billions to own it.
There is also a lesson for founders. Huge AI outcomes do not require building the world’s biggest model. Valuable infrastructure can emerge around model choice, payments, security, evaluation, routing and management.
11. Rillet
Rillet is trying to rebuild one of the least glamorous but most important parts of business software: accounting.
Its argument is that old enterprise resource planning systems were designed mainly to store records. An AI-native financial system should instead understand financial events, automate repetitive work and help accounting teams close their books faster.
Investors are buying into the idea. In August 2026, Rillet raised a $100 million Series C at a $1 billion valuation, bringing total funding above $200 million. TechCrunch reported more than 600 customers and said the company’s annualized revenue had doubled during the previous quarter.
Why Rillet Deserves Attention
Accounting software is not a small niche.
Systems such as NetSuite and Oracle sit deep inside companies and are extremely hard to replace. That creates both the opportunity and the challenge for Rillet.
If AI can reduce manual bookkeeping, reconciliations and month-end work, the value is easy for customers to understand. Finance teams can measure hours saved, faster closes and lower outside-service costs.
But accounting also punishes mistakes. Rillet therefore has to prove that AI-native does not mean less reliable.
Its growing alliances with firms such as EY are worth watching because those relationships can help it enter larger organizations where trust matters as much as technical performance.
12. Mirage
Mirage is the company behind Captions, a widely used AI video product.
In March 2026, Mirage raised $75 million in growth financing, bringing its total funding to more than $175 million. The company said Captions had more than 20 million users and that users had produced over 250 million videos.
The company has expanded beyond basic video editing and is positioning itself more like an AI lab focused on video creation, advertising and marketing.
Why Mirage Fits New York
New York is one of the world’s largest advertising, media, fashion, marketing and creator-economy centers.
That makes AI video more than an interesting technical field here. It is software being built next to a huge potential customer base.
The battle will be difficult. AI video is becoming crowded, and companies with enormous computing budgets are improving their products quickly.
Mirage’s best path may therefore be workflow rather than raw model performance alone. If it understands how marketers actually need to produce, edit, approve, test and distribute videos at scale, it can create value that goes beyond generating a pretty clip.
13. Hebbia
Hebbia became one of the early names associated with using generative AI for serious document analysis.
Its Matrix product allowed users to work across large amounts of information in a structured grid, making the software attractive to financial and legal teams dealing with dense research.
Hebbia raised a $130 million Series B in 2024 at roughly a $700 million valuation. At the time, TechCrunch reported that the company had about $13 million in annual recurring revenue and was profitable, although Hebbia’s CEO declined to confirm those revenue details.
In August 2026, Hebbia launched a major update called Matrix 2.0 along with an assistant named Max. Business Insider reported that pilot users increased actions roughly tenfold while daily engagement tripled.
Why 2026 Is a Test for Hebbia
Hebbia is interesting partly because it now faces the problem every strong early mover eventually faces: competitors copied parts of the idea.
AI research tools for finance and law have become crowded.
That makes Hebbia’s next chapter strategically important. Can it turn an early product insight into a lasting platform, or will features that once looked special become normal parts of enterprise AI?
For buyers, this competition is healthy. Rogo, Hebbia and other companies pushing into overlapping knowledge-work markets should lead to better products and more pressure to prove measurable results.
14. Tennr
Tennr attacks a healthcare problem that does not sound exciting until you understand how expensive it is.
When one medical provider refers a patient to another, documents arrive through many channels and often need to be reviewed, entered, checked and routed manually. Missing information can create delays or cause patients to disappear from the process.
Tennr builds models and workflow software specifically for this referral system.
The New York company raised a $101 million Series C in June 2025. It said its platform had helped process millions of patients across hundreds of healthcare providers and that revenue had more than tripled since the previous funding round.
Fortune reported that the round valued Tennr at $605 million and that the company was founded in 2021.
Why Tennr Is a Good Example of Practical AI
Healthcare has plenty of AI hype around diagnosis and drug discovery.
Tennr’s opportunity is much less dramatic but potentially easier to turn into clear business value.
Administrative work is expensive. Delays hurt both providers and patients. Referral teams already know where the pain is.
That gives Tennr a strong starting point: it does not need to convince the market that the problem exists. It needs to prove that its technology can handle messy healthcare information accurately enough to remove manual work.
15. Pinecone
Many useful AI applications need to answer questions using information that was not part of a model’s original training data.
That is where technologies such as vector databases became important.
New York-based Pinecone built one of the best-known products in this category. Its database helps applications find information based on meaning and similarity, making it useful for search, recommendation systems and AI applications that need access to a company’s own knowledge.
Pinecone said in 2025 that it had more than 5,000 customers and had raised $138 million.
Why Pinecone Is Worth Watching Even After the First AI Hype Wave
Infrastructure markets change quickly.
Features that once required a separate company can become part of cloud platforms, databases or model providers. Pinecone therefore cannot rely only on being early to vector search.
Its opportunity is to keep moving up the stack and make it easier for businesses to build accurate AI systems connected to their own information.
This matters because companies are learning that a good general model is only one part of an enterprise AI product. Data access, permissions, retrieval quality, security and monitoring often decide whether the system becomes useful in production.
Pinecone sits close to that problem.
16. Viam
Viam brings AI into machines rather than keeping it inside a browser.
Founded by MongoDB cofounder Eliot Horowitz, the New York company provides an engineering platform connecting software, data, devices, automation and AI.
Viam raised $30 million in Series C funding in March 2025, bringing total funding to $117 million. The company describes its mission around applying data and AI to the physical world.
Why Physical AI Could Become a Bigger New York Story
Generative AI began with text because the internet provided huge amounts of training data.
Machines create harder problems. Robots need to understand cameras, sensors, movement, physical limits and unpredictable environments.
New York has started building more support around this field. Tech:NYC and Accenture point to physical AI across areas such as drones, sensors and robotics, while New York institutions and organizations are building related research and talent networks.
Viam can benefit if more companies want a common software layer connecting those physical systems.
The important question is whether physical AI moves from exciting demonstrations into large repeatable business deployments. If that happens, infrastructure companies around the machines could become extremely valuable.
17. Nayya
Employee benefits are filled with decisions people rarely feel prepared to make.
Workers have to compare health plans, supplemental benefits, financial programs and retirement choices, often using documents full of unfamiliar terms.
Nayya uses AI and employee data to help people make those decisions.
Forbes reported in February 2026 that the New York company had raised $130 million, generated more than $30 million in annual revenue during 2025, and reached nearly one million users, many through large partners.
Why Nayya Represents a Different Kind of AI Opportunity
Nayya is not trying to impress users with a general-purpose assistant.
It is trying to improve a high-friction decision that already happens inside almost every large employer.
That gives it useful distribution possibilities through benefits providers, payroll companies, employers and financial platforms.
The company’s 2025 acquisition of Northstar also pushed it further into financial wellness. The interesting long-term question is whether Nayya can become an intelligence layer sitting across health benefits, personal finance and employee decision-making rather than remaining a tool used mainly during benefits enrollment.
18. Reserv
Reserv is taking AI into insurance claims.
The company operates as a third-party administrator, or TPA, which means it actually helps insurers manage claims rather than simply selling them another software dashboard.
That distinction matters because Reserv can design technology around the work itself.
In May 2026, Reserv announced a $125 million Series C led by KKR. The company said it served nearly 200 insurers, corporate captives, managing general agents and brokers.
Other reporting put Reserv at approximately $100 million in annual recurring revenue and said the company wants to expand its annual claims capacity from roughly 500,000 complex claims to 30 million within four years.
Why Reserv’s Model Is Important
One of the biggest questions around AI agents is whether startups should sell software to existing workers or rebuild the service itself around AI.
Reserv leans toward the second approach.
That model can create more value because the company controls more of the workflow. It can also create more operational risk because insurance claims involve money, regulation and customer trust.
If Reserv can combine human adjusters with AI in a way that improves speed without lowering quality, it could become a strong example of how traditional service businesses are rebuilt in the AI era.
19. TollBit
The rise of generative AI created a basic economic fight.
AI companies want huge amounts of useful internet content. Publishers spend money producing that content and increasingly want to control or charge for how automated systems use it.
TollBit is trying to create infrastructure between those two sides.
The New York startup raised a $24 million Series A in October 2024, following almost $7 million in earlier seed financing. At the time, the company said more than 200 publisher sites were being onboarded.
Its system lets content owners identify AI bot traffic and create paid access arrangements.
Why TollBit Could Become More Important Than Its Funding Suggests
TollBit is one of the smallest companies on this list by disclosed capital.
That is exactly why funding alone is a poor way to rank AI businesses.
The open web is being reshaped by AI search, answer engines and agents. If fewer people visit original websites because an AI system gives them the answer directly, publishers need new economic models.
Some large publishers will sign direct licensing contracts with model companies. Thousands of smaller sites cannot negotiate individual agreements with every AI platform.
A common payment and permission layer could help solve that coordination problem.
The market is still early, but the problem is getting larger rather than smaller.
20. Arthur AI
As AI systems become more powerful, businesses need to know whether those systems actually work.
That sounds obvious. In practice, evaluation is one of the hardest problems in enterprise AI.
A model can give a strong answer today and fail on a slightly different task tomorrow. Agents add more complexity because they may use tools, call other models, take actions and complete several steps before producing a result.
New York-based Arthur develops tools for monitoring, evaluating and governing AI systems. The company raised a $42 million Series B in 2022, taking disclosed funding above $60 million.
Arthur has since expanded its products for generative and agentic AI. In 2025 it released an open-source real-time evaluation engine and later added monitoring and tracing designed for AI agents.
Why Arthur Could Benefit From the Agent Boom
Every successful AI application creates a second market around making that application dependable.
Companies need to know which prompts fail, where agents make mistakes, what private information is being exposed and whether an updated model improves or damages performance.
That problem becomes more serious as agents move from giving advice to taking actions.
Arthur has been working on AI reliability since before generative AI became mainstream. Its challenge now is turning that experience into an advantage while cloud providers and newer startups move aggressively into the same market.
What Makes New York Different From Other AI Hubs?
The obvious comparison is San Francisco.
The Bay Area remains extraordinarily strong in frontier-model development, research talent, venture capital and infrastructure. New York does not need to pretend otherwise to make a credible case for itself.
NYC’s advantage is different.
New York Has Customers Within the Ecosystem
A financial AI founder can talk to bankers, hedge funds, private equity firms, insurers and asset managers without leaving the city.
A legal AI company can build relationships with major law firms and regulated businesses. A healthcare startup has access to enormous health systems. A media AI company is surrounded by publishers, advertising agencies, entertainment businesses and brands.
This short distance between the builder and the buyer can speed up product development.
The strongest AI products often require more than model engineering. Teams need to understand why customers do a task a certain way, which data they can use, where legal limits exist and how an AI tool needs to fit existing software.
That kind of industry knowledge is one of New York’s deepest assets.
New York Is Developing Both Applied and Frontier AI
The phrase “applied AI” remains a good way to describe NYC’s main advantage, but our research suggests that it should no longer be interpreted as “New York only builds applications.”
Reflection is building frontier open models. Runway is building world models. General Intuition is training action models. Flourish is studying biological intelligence. Hugging Face has become global infrastructure for open AI development.
That creates a healthier ecosystem.
Applied companies create customers and revenue. Infrastructure companies create tools that other startups can use. Research labs attract specialized scientists. Successful exits return employees and capital to the local startup market.
Those layers can reinforce one another.
Talent Demand Is Becoming Broader
AI hiring is not only about machine-learning researchers.
New York companies increasingly need software engineers, product managers, designers, finance experts, lawyers, healthcare operators, salespeople and implementation teams who understand how to bring AI into real businesses.
The 120% year-over-year increase in AI-related job postings reported by Tech:NYC and Accenture supports that shift. Their research also found that 84% of employers surveyed had already upskilled existing workers into AI roles.
That matters because it means AI growth does not depend only on importing a small number of elite researchers. It can also pull existing New York industry talent into new kinds of technology work.
What NYC Businesses Should Learn From These Companies
The biggest mistake a business can make after reading a list like this is to say, “We need AI,” and then start buying tools.
Start with the workflow instead.
Look for Expensive Repetitive Work
Rogo is attractive to finance firms because junior bankers spend huge amounts of expensive time collecting and organizing information.
EliseAI attacks communication and administrative work at housing and healthcare organizations. Tennr focuses on messy referral processes. Rillet attacks manual accounting work. Reserv targets claims operations.
The pattern is consistent.
The best AI opportunities often sit where a company has a process that is repetitive, expensive, data-heavy and easy to measure.
That gives leaders a much better starting point than asking employees to brainstorm possible chatbot uses.
Use This Decision Table Before Buying AI
| Business Problem | Type of AI to Evaluate | NYC Companies Showing the Pattern | Metric to Track |
|---|---|---|---|
| Staff spend hours researching documents | Enterprise research AI | AlphaSense, Rogo, Hebbia | Hours saved per project |
| Customers wait for basic responses | Vertical AI agents | EliseAI | Response time and resolution rate |
| Accounting close takes too long | AI-native finance software | Rillet | Days to close books |
| Healthcare referrals are delayed | Workflow AI | Tennr | Time from referral to scheduling |
| Claims require heavy manual handling | AI-enabled services | Reserv | Cost and time per claim |
| Legal review slows operations | Compliance agents | Norm AI | Review time and exception rate |
| AI apps need access to internal data | AI infrastructure | Pinecone | Retrieval accuracy and answer quality |
| Teams use many model providers | Model-routing infrastructure | OpenRouter | Cost, latency and model quality |
| AI systems fail unpredictably | Evaluation/governance | Arthur AI | Failure rate before and after controls |
| Physical equipment needs smarter software | Physical AI | Viam, General Intuition | Automation rate and intervention rate |
The final column is the most important.
Do not measure an AI project by how often employees open the application. Measure whether it improves the work.
What Founders Can Learn From NYC’s Strongest AI Startups
The data suggests that New York founders do not need to chase the same opportunity as every other AI founder.
In fact, doing so may waste the city’s strongest advantage.
Build Where NYC Has Unfair Access
Someone building software for private equity can spend years trying to understand how deal teams work.
A founder who previously worked in private equity already knows.
The same is true for insurance, fashion, law, healthcare, advertising, real estate and media.
Founders should ask themselves a simple question: What can we understand in New York that a generic AI team somewhere else would struggle to learn?
That knowledge can become part of the moat.
Proprietary Workflow Data May Matter More Than Another Model
General Intuition’s story is useful here.
Its advantage is not simply that its researchers know how to train AI. The company started with access to gameplay plus the human action data connected to that gameplay.
Other vertical startups can create similar advantages on a smaller scale.
A healthcare company can learn from referral patterns. An insurance company can build knowledge from claims. A financial platform can learn how professionals complete complex research. A property platform can learn from millions of resident conversations.
Models increasingly become available to everyone.
Unique data generated by real workflows is much harder to copy.
Sell an Outcome, Not “AI”
Businesses are getting tired of AI demos.
They want lower costs, faster work, more revenue, fewer errors or a better customer experience.
The companies in this list that appear strongest commercially tend to have a clear answer to the question, “What gets better after I buy this?”
That should influence everything from the product roadmap to the website headline.
What Investors Should Watch in NYC AI During the Rest of 2026
Funding volume is impressive, but investors should be careful about confusing financing momentum with business quality.
Our analysis found that roughly two-thirds of the disclosed funding in this 20-company group sits in only five companies. That concentration means the headline market can look stronger than the experience of the average startup.
Several other signals may be more useful.
Watch Revenue Growth After Large Rounds
Large financing creates expectations.
AlphaSense has a strong benchmark because it paired its $350 million raise with more than $600 million in ARR. EliseAI paired its 2025 round with substantial revenue and customer growth. Rillet’s newest round arrived while customer count and annualized revenue were rising quickly.
For younger frontier labs, revenue may come later. In those cases, investors should look harder at technical progress, computing access, developer adoption and whether the product creates a path toward a defensible market.
Watch Whether AI Agents Move From Assistants to Operators
The biggest enterprise shift may come when AI stops waiting for questions.
AlphaSense’s SuperAnalyst, Rogo’s Felix and products from Norm are all moving toward systems that complete multi-step work.
That can create much more value than chat alone, but it also creates more risk.
The companies that solve approval rules, permissions, monitoring and human handoffs may therefore become just as important as the companies building agents.
Watch Physical AI
General Intuition, Viam and New York’s expanding robotics ecosystem point toward another possible growth area.
Text AI can already produce useful work because words are digital. Physical AI has to handle the unpredictable real world.
If general models make robots easier to train and deploy, enormous industries could be affected. Warehouses, construction, logistics, manufacturing, retail, transportation and building operations all contain physical tasks that software alone cannot perform.
New York does not yet dominate this market. It now has enough activity that it deserves close attention.
The Biggest Risks Facing New York’s AI Boom
Strong growth does not remove risk.
It often creates new kinds of it.
AI Funding Can Run Ahead of AI Economics
A $2 billion funding round does not prove that a company has built a $20 billion business.
Private AI valuations increasingly reflect expectations about markets that do not fully exist yet.
Some companies on this list will almost certainly become much larger. Others may struggle to turn technical progress into durable economics.
That is normal in a technology wave this large.
Businesses evaluating vendors should therefore look beyond valuation. Ask about customers, financial durability, model costs, implementation requirements, security and how difficult it would be to move your data if the vendor disappeared.
Model Improvements Can Destroy Product Advantages
An AI startup can spend a year building a feature that a foundation-model provider adds to its API next month.
That is one of the largest risks facing application companies.
The strongest defense is usually not another thin interface around a model. It is ownership of workflow, trusted data, customer relationships, deep integration or a product that becomes operationally difficult to replace.
Rogo wants to sit inside financial work. EliseAI sits inside housing and healthcare operations. Reserv handles claims. Rillet wants to become the accounting system itself.
Those positions are harder to wipe out with one model update.
Regulation Will Matter More
New York’s AI industry is developing inside one of the world’s most regulated business environments.
That can slow companies down, but it can also create opportunity.
Tech:NYC reported that more than 85 AI-related bills were introduced during New York’s 2025 state legislative session. At the same time, businesses are increasingly concerned with privacy, model safety and responsible use.
Companies such as Norm and Arthur exist partly because AI cannot move deeper into important workflows without stronger controls.
For NYC, regulation may therefore become both a constraint and a new technology market.
The Larger Story: New York Is Becoming an AI City, Not Just a Tech City
There is a difference.
A city can contain technology companies without technology changing how its main industries operate.
AI is starting to move through New York differently.
Wall Street is adopting AI research tools. Property companies are using agents for resident and leasing operations. Healthcare companies are automating paperwork. Accountants are testing AI-native financial systems. Publishers are trying to build a new economic relationship with AI crawlers. Researchers in Manhattan and Brooklyn are building frontier models and studying alternatives to today’s AI architectures.
The city’s public and academic infrastructure is moving too. NYCEDC’s AI strategy, the NYC AI Nexus, Empire AI, Cornell Tech, Columbia, NYU, CUNY and other programs are creating additional links between founders, researchers, students and businesses. NYCEDC’s AI Nexus initiative alone was designed to accelerate collaboration between local startups and businesses adopting applied AI.
This creates a powerful feedback loop.
Industries create problems worth solving. Startups build products around those problems. Customers generate revenue and specialized data. Revenue attracts investors. Successful companies attract workers. Workers start more companies.
New York already has most pieces required for that loop.
The question is whether they continue compounding.
Frequently Asked Questions About AI Companies in NYC
How many AI companies are in New York City?
NYCEDC says New York City has more than 2,000 AI startups, while the wider technology ecosystem includes more than 25,000 startups. The New York metro area also has more than 40,000 workers with AI-related skills.
The exact number depends on how “AI company” is defined. Thousands of normal software companies now use AI, but we would not classify every company offering an AI-powered feature as an AI startup. For this article, AI had to be central to the product, infrastructure or research.
Is NYC becoming a major AI hub?
Yes. The stronger question now is what kind of AI hub it will become.
New York’s main advantage remains applied AI because the city has unusually dense concentrations of finance, media, advertising, healthcare, law, insurance, real estate and enterprise customers. However, the rise of Reflection AI, Flourish, General Intuition and Runway shows that more frontier and research-heavy AI work is now happening in New York as well.
Which NYC AI companies have raised the most money?
Among the companies in our sample, Reflection AI is far ahead based on reported funding close to $2.6 billion. AlphaSense has raised well over $1 billion, while Runway has raised approximately $860 million.
Funding should not be treated as a direct measure of quality. Research labs require far more capital than many software businesses, so comparing their funding totals without considering their business models can be misleading.
What industries are strongest for AI companies in New York?
Our research suggests that enterprise knowledge work is particularly strong. Finance, legal services, accounting and market intelligence alone represent five of the 20 companies in our sample.
Healthcare, housing, insurance and employee benefits are another major group. Creative AI remains important because of New York’s media and advertising industries, while physical AI is emerging as a smaller but fast-moving part of the ecosystem.
Which NYC AI company should businesses watch most closely?
There is no single right answer because the companies solve very different problems.
Large enterprises should pay close attention to AlphaSense because it offers one of the clearest examples of an AI company reaching large-scale recurring enterprise revenue. Financial firms should follow Rogo and Hebbia. Housing and healthcare operators should watch EliseAI. Finance teams should track Rillet. Developers should watch Pinecone and the changes around OpenRouter and Hugging Face. Companies deploying agents should keep an eye on Arthur and Norm because evaluation, governance and legal controls are becoming more important.
Is New York likely to overtake San Francisco in AI?
That is probably the wrong way to measure success.
San Francisco has an exceptional concentration of frontier AI laboratories, engineers, investors and computing companies. New York has a different advantage: it connects technical talent to an enormous range of industries that can become AI customers.
Both cities can grow at the same time.
The more useful question is whether New York becomes the best place to build AI for industries where specialized knowledge, customer access, trust and regulation matter. Current evidence suggests it has a strong chance.
Final Thoughts
The most important takeaway from New York’s AI market in 2026 is not that startup funding is rising.
It is that the type of company being built here is changing.
New York still excels at turning artificial intelligence into tools for finance, real estate, healthcare, insurance, media and professional work. That strength is becoming deeper as companies such as AlphaSense, EliseAI, Rogo and Rillet build AI directly into expensive business processes.
At the same time, a second layer is emerging. Reflection is building frontier open AI. Runway and General Intuition are working on systems that understand environments and actions. Flourish is investigating whether lessons from biological brains can produce a different kind of artificial intelligence. Hugging Face has made Brooklyn home to one of the world’s most important open AI platforms.
That combination is what makes NYC’s position interesting.
Fourteen of the 20 companies in our sample—70%—are applied or physical AI businesses, while a smaller group of core AI and infrastructure companies attracts a much larger share of the capital. Across the entire group, we estimate at least $8.3 billion in disclosed or credibly reported funding, and ten clearly verified 2026 rounds alone account for another $2.178 billion of financing activity.
Those numbers will change. Some of the companies will be acquired. Some will grow far faster than expected. Others will discover that impressive technology is easier to build than an enduring business.
But the direction is difficult to miss.
New York already had the customers, capital, universities, talent and industries needed to become an AI powerhouse. It is now developing the model labs, infrastructure companies and AI-native businesses to connect those pieces together.
For founders, that creates opportunity.
For NYC businesses, it creates a growing local market of serious AI vendors.
For workers, it means AI careers will increasingly exist far beyond traditional software engineering.
And for anyone watching where the next generation of major artificial intelligence companies will come from, New York City now belongs near the top of the list.
