The State Of The NYC Tech Ecosystem

Matt Turck has penned a “State Of The City” post about where the NYC tech ecosystem is right now. I get asked this question all the time and I haven’t been doing a great job of answering it. I will use some of Matt’s work the next time that happens.

Here’s some of my favorite points from Matt’s post. If you live and work in the NYC tech ecosystem, or care about it, you should go read the whole thing.

NYC as a leading AI Center:

The New York data and AI community, in particular, keeps getting stronger.  Facebook’s AI department is anchored in New York by Yann LeCun, one of the fathers of deep learning.  IBM Watson’s global headquarter is in NYC. When Slack decided to ramp up its effort in data, it hired NYC-based Noah Weiss, former VP of Product at Foursquare, to head its Search Learning and Intelligence Group.   NYU has a strong Center for Data Science (also started by LeCun).  Ron Brachman, the new director of the Technion-Cornell Insititute, is an internationally recognized authority on artificial intelligence.  Columbia has a Data Science Institute. NYC has many data startups, prominent data scientists and great communities (such as our very own Data Driven NYC!).


NYC as a home to “deep tech”:

Finally, one trend I’m personally particularly excited about: the emergence of deep tech startups in New York.   By “deep tech”, I mean startups focusing on solving hard technical problems, either in infrastructure or applications – the type of companies where virtually every early employee is an engineer (or a data scientist).

For a long time, MongoDB was pretty much the lone deep tech startup in NYC.  There are many more now.  A few of those are in my portfolio at FirstMark:  ActionIQ, Cockroach Labs, HyperScience and   But there’s a lot of others, big and small, including for example: 1010Data (Advance), BetterCloud, Clarifai, Datadog, Dataminr, Dextro, Digital Ocean, Enigma, Geometric Intelligence, Jethro, Placemeter, Security ScoreCard, SiSense, Syncsort or YHat – and a few others.


The Diversification and Broadening of NYC’s Tech Ecosystem:

One way of thinking about New York’s tech history is one of gradual layers, perhaps something like this:

  • 1995-2001: NYC 1.0, lots of ad tech (Doubleclick) and media (TheStreet)
  • 2001-2004: Nuclear winter
  • 2004-2011: NYC 2.0, a new layer emerges around commerce (Etsy, Gilt) and social (Delicious, Tumblr, Foursquare), on top of adtech (Admeld) and media
  • 2012-present: NYC 3.0 – in addition to the above, just about every type of technology covering just about every industry

Certainly, the areas that put NYC on the map in the first place continue to be strong.  New York is the epicenter of the redefinition of media (Buzzfeed, Vice, Business Insider, Mic, Mashable, Bustle, etc.), and also home to many great companies in adtech (AppNexus, Tapad, Mediamath, Moat, YieldMo, etc.), marketing (Outbrain, Taboola, etc) and commerce (BarkBox, Birchbox, Harry’s, Warby Parker, etc.).

But New York has seen explosive entrepreneurial activity across a much broader cross-section of verticals and horizontals, including for example:

  • Fintech: Betterment, IEX, Fundera, Bond, Orchard, Bread
  • Health: Oscar, Flatiron Health, ZocDoc, Hometeam, Recombine, CellMatix, BioDigital, ZipDrug
  • Education: General Assembly, Schoology, Knewton, Skillshare, Flatiron School, Codecademy
  • Real estate: WeWork, HighTower, Compass, Common, Reonomy
  • Enterprise SaaS: InVision, NewsCred, Sprinklr, Namely, JustWorks, Greenhouse, Mark43
  • Commerce infrastructure: Bluecore, Custora, Welcome Commerce
  • Marketplaces: Kickstarter, Vroom, 1stdibs
  • On Demand: Handy, Via, Managed by Q, Hello Alfred
  • Food: Blue Apron, Plated
  • IoT/Hardware: littleBits, Canary, Peloton, Shapeways, SOLS, Estimote, Dash, GoTenna, Raden, Ringly, Augury, Drone Racing League
  • AR/VR/3D: Sketchfab, Floored



I like the NYC 3.0 moniker. It’s a very different place to start and invest in tech companies than it was even five years ago. Bigger, deeper, broader, and scaling nicely. Just like the companies themselves.

#entrepreneurship#NYC#VC & Technology

Comments (Archived):

  1. Joe Lazarus

    Good read. I’m curious, which NY startups are the AVC community most optimistic about in terms of future valuation growth?

    1. Ana Milicevic

      I think MediaMath has the potential to be as influential and as big a commercial success as DoubleClick was.The whole IoT category from Matt’s article is incredibly exciting – especially JewelBots, littleBits and Ringly.ZipDrug has phenomenal leadership and a great solution for consumers (as long as they can avoid getting targeted by the pharma cartel the way PillPack has been).

    2. falicon

      Each of our own of course! 😉

  2. pointsnfigures

    and no where do they call it Silicon anything. Nice.

    1. Jess Bachman

      Not to burst your bubble or anything, but Silicon Alley has been around for a while and is still used.http://www.businessinsider….

      1. fredwilson


    2. Vasudev Ram

      Gold (or Platinum) anything (Alley, other) would probably be a better name. NYC, after all …

    3. Ana Milicevic

      If we do have to have a Silicon- moniker I’d prefer it to be Silicon Empire. Alley doesn’t really do justice to NY’s ambitions and we are technically in the Empire state.

  3. awaldstein

    Great read.Terrific to learn new things about the world you work in.

  4. JLM

    .Very informative.Well played.JLMwww.themusingsofthebigredca…

  5. Ana Milicevic

    We’re growing up!Nuclear winter sure wasn’t fun. I’ve always liked the cross-pollination of industries we have here vs on the West Coast.

  6. Joshua Acosta

    Id be curious to see how the diversity numbers look in NYC 3.0 since I’ll be joining a team at one of these companies soon.

    1. fredwilson

      getting better quickly

  7. William Mougayar

    The diversification of sectors, industries, and technologies was very evident in that report, and that’s a very good thing.

  8. sigmaalgebra

    That’s a LOT of companies!There seems to have been a lot of activity in the last few years — amazing.In particular, that’s a lot of people believing in artificial intelligence (AI).Sorry, guys, I tried AI: I thought it was silly. Then I took one of our core problems and just did some applied math and totally knocked our AI solution or anything like an AI solution into the nickel seats. Yes, I published the work.AI was just fumbling around with no definite assumptions by the work, conclusions of the work, or properties of the results of the computations. What I did in math on the same problem was logically, rationally, and mathematically rock solid — with the results, you know in very solid terms just what the heck you’ve got.Broadly, AI is step backwards from good, traditional approaches in applied math, physical science, and engineering.I still think AI is silly — a lot of hype and the approach of all the rest essentially inferior to what we’ve had going all the way back to, say, Newton.To borrow an NYT writing device, of course AI is nonsense. The question is, how is anyone in NYC paying the rent from revenue from AI?Also, the AI people suffer the same problem as computer science more generally — they don’t know how to write up their work. The writing standards in the better known texts in early grad school pure math is much better. Of course, polished beyond belief is Bourbaki.It’s easy enough to learn those writing standards — just study the texts and/or take some good courses from them. Such authors? Sure: D. Bertsekas, L. Breiman, E. Coddington, J. Doob, N. Dunford, U. Grenander, J. Hale, P. Halmos, I. Herstein, I. Karatzas, A. Kolmogorov, D. Luenberger, J. Neveu, H. Royden, W. Rudin, J. Schwartz, S. Shreve, G. Simmons, J. Tukey, etc. So, just pure math in algebra and analysis. For some engineering? Look at the flurry of writing on the fast Fourier transform.I’ve heard of Yann LeCun only recently and only for hisIf you do research in isolation, the quality goes down. Which is why military research sucks.easy to find at Google but in particular at…Very, very sorry, Yann, but you are wildly wrong on both points.

    1. Ronnie Rendel

      AI is a collective term for a lot of different things. My experience has been that for classification problems (for example), ML algorithms allow us to account for many more cases and automatically build more accurate metadata then manually writing algorithms that account for every specific case we expect.So it’s all about what type of problem are you solving.

      1. sigmaalgebra

        Right, you are describing statistics done badly. So, it’s empirical curve fitting.So, what are the rates of false positives and false negatives? How close are the results to the classic Neyman-Pearson result? How does that work compare with classic discriminate analysis in multivariate statistics? What the heck are the assumptions? When the assumptions hold, what are we guaranteed about the results? Is some relevant performance measure optimized; if so, then what is it?I know; I know; we use training data to fit a model and then test that fit on test data. We make some attempt to avoid over-fitting, and, finally, there is some math in that.But we are not at all clear about what we are assuming and can’t be clear about what we’ve got.So, by analogy, instead of Newton using his law of gravity and second law of motion to describe and predict the motions of the planets and much more, we are stuck 2000 or years earlier with Ptolemy and his empirical fitting to observed data with his epicycles. Of course, Ptolemy’s work didn’t fit very well. And Ptolemy had no idea of generalizing to the moons of Jupiter, asteroids, comets, a neutron star orbiting a black hole, how to design a rocket to put 2000 pounds in a geo-stationary orbit, how to get the rocket there for minimum fuel (deterministic optimal control theory, e.g., calculus of variations, e.g., some of what Newton did, e.g., on the Brachistochrone Problem) all of which follows from what Newton did or with Einstein’s general relativity corrections.There are some places for such uses of, say, sigmoid functions for approximating data, e.g., as D. Bertsekas did for the optimal value functions in stochastic dynamic programming, but there the only claim was a good, computationally convenient, fit to some specific data with nothing said about fits to more data. It was essentially a case of data compression.So, you mentioned a case of cookbook statistics done badly.So, we don’t have applied math. We don’t even really have much in statistics. We don’t have science. All we have is some empirical fitting to data that may or may not mean anything in the future.To say something solid about the future, we need some assumptions. E.g., we have the Pythagorean theorem and know that it works for all right triangles, not just the ones in some training set. And we know the key assumption — it’s a right triangle. Given that, the theorem will work on Mars, and we know that before going there.E.g., when Bell was sizing the voice long distance network for Mother’s Day, they had tools much more powerful than some empirical training data. They had the Poisson process and the renewal theorem.When JPL is trying to navigate to Pluto, they don’t have to use some training data. Instead they use mostly just the law of gravity and Newton’s second law of motion with maybe some correction for the pressure of sunlight and the solar wind.And that empirical curve fitting is called artificial intelligence. Not thrilled about artificial and prefer something solid and genuine. So, what we’ve got is solid and genuine trivia plus a lot of hype.We are in line soon for another “AI winter”.AI is an old word play scam going back to at least the publicity about IBM’s early computers that were described as “gigantic electronic human brains”.So, each two decades or so, someone dusts off this word play scam, writes some code to do something on some problem, pumps out a lot of hype, gets headlines about artificial intelligence, and the media get back to their anxiety stories about robots taking over all the jobs and the planet and gets more headlines, eyeballs, and ad revenue.Then, like teenage girl fashion frocks, soon enough the headlines are off to something else, and there is yet another AI winter. Then, sure, wait another 20 years and do it again.Trivia, hype, word play scam — nothing I’d want anything to do with.For investing, if the AI is from a public company, then watch the rumors 10 times a second 24 x 7 with running shoes on and be prepared to run to the exits ASAP. For an illiquid investment, better have some darned good reasons to invest other than artificial intelligence.

    2. Twain Twain

      Link to your published paper, please. Thanks.I’ve been at Data Science Summit in SF where some maths was shared. Current AI is narrowly defined for simple reasons: objective function is to increase ads, reduce churn, retain interaction on startup’s assets.Academic research is likewise oriented because the techcos pay for it and the professorships.

      1. sigmaalgebra

        I’m anonymous on Disqus.Ah, right! A startup opportunity!If you can give me an e-mail address, then I will send you the reference and, better yet, a PDF of the paper.Prerequisites include measure theory, probability theory based on measure theory, some group theory, and some more elementary topics. The thing is based on some symmetry — in a crucial sense, the input observed data is not symmetrical but the probability distribution of the data is symmetrical, and that’s enough. The symmetry, then, leads to a group of transformations, and at the end just sum over the group. Yup, it’s theorem-proof type applied math.So, net, the paper is some slightly novel derivations in applied probability. Much of the big step past the AI we were doing is that I attacked the problem via probability instead of just intuitive heuristics. When can do probability derivations, they commonly totally knock the socks off anything via intuitive heuristics.So, it’s not just some curve fitting or some heuristic thing-y. Instead, know what the assumptions are and, given the assumptions, have some guarantees about the results.AI is narrowly defined for simple reasons: objective function is to increase ads, reduce churn, retain interaction on startup’s assets.Terrific — they actually have discovered objective functions. WOW! Maybe in another few years they will also be out of diapers!So, can they actually formulate their optimization problem? I wonder!Ah, optimization! Wait until they stumble on the constraint qualifications for the Kuhn-Tucker necessary conditions for optimality — will lose 80% of the class!For my startup, once it gets into real revenue, I will need to do some version of ad targeting. The math I have outlined is for a cleanly specified problem, and I have a solution that in an important sense is optimal, that is, best possible. And I don’t call it AI — it’s just some applied math.It’s fair enough to call my math original, but there are some prerequisites. If my work is good, then it’s because of the shoulders I stood on. Some of the prerequisites are some just astounding results — wouldn’t believe that any such things could be true, but they are — in, say, W. Rudin, Real and Complex Analysis.

        1. Twain Twain

          twainventures [at] gmail

          1. sigmaalgebra


    3. Twain Twain

      Yann Le Cun and the guys featured in this video are today’s “Fathers of AI”. Their students are the ones creating a number of the AI startups listed:*…I pointed out to Professor Domingos at Data Science Summit yesterday that the AI brain is an inversion of the human brain and, therefore, it can’t be claimed in any way that AI has “human-like” intelligence.He conceded I’m right.

      1. Twain Twain

        It’s so obvious why AI can’t understand the meaning of human languages and values when we breakdown what AI can do and what it focuses on (autistic mathematical logic and probability).

      2. sigmaalgebra

        > an inversion of the human brainIf I knew what inversion meant in that context, then I could think about that claim.

          1. sigmaalgebra

            Nice diagram.First, I don’t know how the human brain works!Second, I have some wild guesses about some of how, maybe, if my wild guesses had a good track record, which they don’t, real artificial intelligence might be programmed.Third, for what has been programmed and especially what might be programmed, I’m not sure the diagram has to hold. E.g., I see no fundamental reason one might not start with emotions and move forward from there.E.g., my startup takes in user data on what usually will be from their feelings, emotions, etc. and pushes that data through some math to give some results they stand to like. That is a long way from anything to be called intelligent, but it might show that we can have computers start with emotions and work forward from there.That is, we really can if we just figure out how, and I have to believe that there is no solid reason we can’t do that.Computers start with emotions? Gee, I have a computer! At Hacker News about an hour ago I saw an article about violins, new ones versus old Stradivari ones. That led me toJanine Jansen, Jules Massenet,”Meditation”, Thaïs…and that is awash in emotion: She plays it like she really, really get’s it, but it did come through my computer!That URL should be in the initial data for my startup!

  9. Tom Labus

    The long knives were out (and still are) for IEX. It’s great to see them become an exchange.

    1. pointsnfigures

      Agree. SEC and CFTC make it really hard (and expensive) to become a full fledged exchange.

    2. JamesHRH

      Deeply proud of the CDN roots of IEX & feel that the pragmatic but tough minded ethos of the company reflects the best of Canadian values.

  10. William Mougayar

    Is there still an interest in “Coming Up With A Better Name For NYC’s Tech Community?”…

    1. fredwilson

      do we need one?

      1. William Mougayar

        Only if one comes up with a really good one.Otherwise the city name is good.

        1. awaldstein

          Strongly advise that a name is not needed at all.

          1. Joe Cardillo

            Naming an ecosystem seems pointless to me…that’s like trying to control it, and usually about 20 mins after you do it things are already evolving.

          2. awaldstein

            Branding an ecosystem does have value as ecosytems are cultural entities as much as fiscal.There already is a strong brand here. In fact, few brands are as inclusive, as personal, as personizable as New York. That’s its power and magic.

          3. Joe Cardillo

            I’d agree with that, though I’m not convinced the tech ecosystem needs anything beyond that. On the other hand SV probably carries more power than CA or SF/Bay area at this point.

          4. fredwilson


          5. Jess Bachman

            Well this would be ideal, but you should know that the best way to remove a moniker is to replace it, not forget it. “voids” don’t market very well. If the powers that be really wanted to remove the Silicon Alley moniker, and replace it with a ‘generic’ like NYC Tech, they need to put a lot of effort and attention into putting NYC Tech out there and into peoples minds. It’s not going to happen on it’s own.

          6. awaldstein

            Actually no.It is true that to change a habit you best replace it.It is true that is is often easier and advisable (though not always) to change the definition of a category rather than create a new one.But to consider NYC as a generic brand is not reality. It is an exception certainly and possibly one of the most powerful brands on their is and more powerful as it is many starred whether you are in arts, music and tech.In this case New York as a brand if you can become part of it is as powerful as you can get.Or so I think.

          7. Jess Bachman

            We are on the same page. I didn’t mean NYC as a generic brand, just a generic word.

          8. LE

            Actually no….Strongly advise that a name is not needed at all.You seem very sure of yourself in your statements. Very black and white. The truth is (and you know this) marketing and business is an art, not a science. Saying so absolutely “no” in rebuttal to Jess doesn’t account for the value of branding for the runner ups. NYC alone is a powerful brand name. But that brand name doesn’t extend to tech and certainly not anywhere near the value created by the use out west with “Silicon Valley”.When you are at the top sure you don’t need names. So maybe NYC can stand at the top in certain areas just like LA can. [1] But tech is not one of those so having a good name would help more than being neutral.[1] “LA” stands on it own with entertainment for example as does “Hollywood”.

          9. awaldstein

            I am positive.And though I rarely say this, simply right on this one.Easy to ask questions. Find ambiguities.This is how it should be.

          10. William Mougayar

            i agree, unless a smart name comes up.

        2. Jess Bachman

          New York DigiThe Big Algo….I’ll see myself out..

      2. JamesHRH

        No.The mistake being made here is the belief / assumption that Silicon Valley is an industry moniker – it is not. It is a geographic name.Where are you guys based? Bay Area.The city or the valley? Etc.Where are you guys based? NYCBrooklyn, Uptown, Midtown or SOHO? Etc.Silicon Alley was an answer looking for a question that wasn’t being asked.

  11. pointsnfigures

    Was thinking about this and the Gartner cycle. http://www.marketingteacher… This is a more traditional marketing chart. Wonder if it makes more sense? I don’t like the “disillusionment” thing. A lot of that depends on how long you have been there as well. New people to the ecosystem bring new energy to it, and can shake up disillusionment. If you are disillusioned, important to check your confirmation bias and network with people on each side of you

    1. Amar

      Small nit: what you have above is the technology adoption cycle that was popularized by Geoffrey Moore’s book “Crossing the chasm”…Gartner is more known for creating the hype cycle ->

      1. Vasudev Ram

        Crossing the Chasm was a good book. I read it some years ago. Apart from the main points conveyed in it (which can be found from reviews of it), one interesting point that stuck with me – I forget the exact term Moore used for it, if any – is how having a main product and then a sort of constellation of allied / auxiliary products and/or add-ons around it, can lead to much large sales and success than just the main product alone, and also those parts all act as a synergistic and self-reinforcing whole.

    2. awaldstein

      This is simply not reality as gospel any longer.I worked with Geoff and hired him for my event series for years.This is not how the world works any longer.

      1. pointsnfigures

        How so @SixgillBlog:disqus ? and @amar:disqus, agree. Saw the Gartner cycle and the other heuristic popped into my head.

        1. awaldstein

          Don’t have time for a response today–sorry but this holds many of the ideas that form the answer.

      2. Vasudev Ram

        >This is not how the world works any longer.Can you elaborate on that? I’m not saying you are wrong. Don’t know enough. Just interested, since I had read that book of his some years ago, and at least some parts of it made good sense. What has changed, for example? (Also, I do realize his book may not have been applicable to all areas even then.)

      3. cavepainting

        I think what has changed is a) the dramatic fall-off in the number of new products that do not make it beyond the innovators bucket, and b) for the ones who make it, the pace at which new products move through the adoption cycle has accelerated exponentially. Something like a slack or to be more current – a pokemon go – can go from innovators to mainstream in a matter of days.

  12. Thor Snilsberg

    Great read and post. Ecosystems are great for learning and innovation. Coming from the STEM education side folks like me running STEM nonprofit efforts/interventions should be aware/participating actively in the national STEM ecosystem mapping project. It puts over a decade of research into building collaborative systems that impact STEM education and ultimately the local STEM workforces. was selected in 2015 to participate and the folks at Expand Ed are front and center in leading the effort, http://www.expandedschools….Here is the inaugural newsletter and list of (outdated) events that show the depth and scale of the STEM ecosystem efforts,

  13. Joe Cardillo

    Has anyone written about strongest tech co’s in NYC that have worked laterally? I’d be curious to know which have supported the most growth in other startups

  14. TeddyBeingTeddy

    Interesting you call out Real Estate. Fred what’s your long term view on suburban commercial office space? Do self driving cars increase demand for it, or does virtual meeting technology and globalization eventually kill demand for large suburban office space?I think relatively inexpensive resort-like real estate goes up as people can increasingly work from anywhere. Why live in a concrete jungle when you can live & work with your friends from the beach?

    1. LE

      Impossible to predict but also you have to keep in mind the effect of younger people (at least for now) wanting to work in cities vs. the suburbs. (one of the reasons GE moved to Boston). [1]That said schools are the main driver of why and where people who have kids relocate to. But that’s now that could easily change for any number of reasons.[1]

      1. TeddyBeingTeddy

        Baby boomers and millennials want to move urban. So tax revenue should go up, thus schools downtown should too. I think more than the schools, millennial will want to move back by their parents because to help raise kids. Millennials are too lazy to work very hard now, imagine the shock trying to also raise kids! So likely depends where their parents move (Florida or a big fun city). Should be interesting either way!

        1. LE

          My brother in law and his wife just got pregnant for the first time (millenials, he is much younger). Her parents live outside of Chicago, and her mother is flying in and living with them for some time just to help out with the new baby. They moved out of the city to a nice suburb within driving distance of “the city”. (He needs it for his career). So they had been living the NYC life for a long time with “the lifestyle”. Nice place where they are but first thing is (as newly married people) they go to some bar in the area and were totally shocked at how the crowd wasn’t what they expected or that they were used to in NYC. I don’t think they’ve gone there since. They picked the area they moved to specifically because of the schools. What’s funny is that there neighbor is an ardent gun rights activist. Like they have signs on their house outside wacky, flags and what not. (Given the pricing they couldn’t afford to be in the best place in the town they moved to but apparently can still take advantage of the schools).

          1. TeddyBeingTeddy

            I lived in Chicago 5-6 years ago before moving out west and after living in NYC. So I can speak to this. Based on those facts, they’d probably be best suited for Oak Park or Evanston. Both lively downtowns and the closest to the city. Evanston’s schools aren’t nearly as good and further from the city, but it’s closer to the water and further from the ghetto. If they want something resembling the hipster side of NYC, and don’t want to own a car, they’ll like OP…best downtown in the burbs. If they want the upper east side, and never need to fly for work, they’ll prefer Chicago’s north shore like Evanston. Anything north of Winnetka is way to far, anything west of La Grange is way to far, anything South of the city is a NFW.

          2. realposter

            I have relatives who live in Evanston.. That is true about Evanston’s schools.. However, just as I wrote to the person above you – parents matter more than the school system. When capable parents pull their children out of neighborhoods and cities because of “the poor school system” – all they do is re-enforce the segregation that made public schools fall apart. I know too many past and current examples here in NYC of students who went to public school and are successful. It’s impossible that it is all bad. That goes for both here in NYC and in Evanston (my younger cousin went to public schools there and she now has her master’s degree).

          3. TeddyBeingTeddy

            I agree with you 100%. All comes down to parenting. But if you’re a bad parent, at least a good school can help offset things a bit ; )

          4. realposter

            What’s strange to me is how “enlightened” parents of this generation still automatically have the “have a child – time to move to the suburbs” mentality. Just as crime in the city was dealt with – public schools can be dealt with. In spite of all it’s challenges – city students every year still make it to all of the top school. Even just in the past 2 days I’ve interacted with 3 former public school students in their 20’s. Two were raised in the Bronx and one other. One of them is an intern at our company for the summer. All 3 grew up in immigrant families. One is in their first year residency as a medical doctor – one got a job at a top law firm in Manhattan. Our intern just graduated from Syracuse in business. My point is they all came out just fine in spite of living in supposed “bad” neighborhoods. In each case they had caring parents – who though not “poor” – just didn’t have the wherewithal to move to the suburbs while adjusting to this country. So they made it work and their kids are a success. All have 2 parent homes. When “good people” flee cities – all it does is perpetuate the segregation that destroyed the schools in the 70’s and onward.

    2. realposter

      I don’t think self driving cars will increase demand for it. As long as crime continues to drop – or remains fairly low – those with ambition and drive will continue to seek it out. That only stopped when cities fell apart in the US starting in the 1960’s. It never changed elsewhere in the world… Cities where crime is manageable in the US have boomed. Humans still need that friction to build innovations and creativity. “Necessity is the mother of invention”.In any event – NYC has 14 miles of beach… You just can’t swim all year round. But if you can spend your time swimming then you aren’t building product. It takes a lot of grind work. I read not too long ago that people in the Bay Area work even longer hours than NY’rs.

  15. John Fazzolari

    Biggest weaknesses seem to be lack of early stage angels to support new entrepreneurs and government officials who want to destroy airbnb

  16. Jon Katzur

    I work at one of the “deep tech” companies listed. I’ve found NY to have some unique advantages in enterprise tech specifically. The biggest being the sheer number of industries centralized here. From Media to Fashion to Finance, there are a ton of _customers_ based here. I think that’s a huge competitive advantage in terms of being exposed, on an every day basis, with potential clients and what their real world problems are.As for talent, while prior startup experience is incredibly helpful in many areas, any companies with data engineering / architecture / analytics domains can benefit significantly from Wall Street talent. Companies like Bloomberg and obviously quant firms are fantastic training ground for learning how to deal with data at scale to get real insights.

    1. JamesHRH

      Bingo!In enterprise, access to customers is key.

  17. Kirsten Lambertsen

    Great informative share. Thanks!I’m perplexed by the absence of blockchain startups here. Too early for them? Or are there not enough of them to make NYC a leader in that category?

    1. Joe Cardillo

      Great question, be cool to know the answer to that

      1. Kirsten Lambertsen

        Ah, I see now he does add a disclaimer at the end of the original post:”This post is meant to be a few thoughts on the topic, rather than a proper “State of the Union” of New York tech. There are several interesting themes I haven’t mentioned…”So it can probably just be taken as a reflection of Turck’s personal experience and perspective.

    2. realposter

      I just read an article that the Wall St. banks and IBM were all facilitating them here in NYC. I can’t remember where I read it – but type it in your browser and see what comes up.

  18. Richard

    Not only are your struggling with the NYC 3.0 monicor but you are compounding it with “deep tech”. How about “Tech Liberty” and “API tech” We are a “Tech Liberty Company”, Find the statute of Liberty, we are not far from there. and we are in the “API tech space”

    1. Joe Cardillo

      Googled and there’s no”Deep Tech Capital” venture firm. Yet…

      1. Richard

        Deep tech feels like a meaningless decryption that a politician would use. Deep vs Shallow, pretty sure enterprise technology company covers it and more.

  19. Richard

    Which VC Bucks the 2/20 model first?

    1. Girish Mehta

      As in drop the mgt fee and go only carry ?

      1. Richard


  20. Sebastien Latapie

    Excellent! And a place I’d love to be. It always surprises me at how big health tech is in NYC

  21. creative group

    Contributors:Do you feel The Internet Of Things will be assisted by the FCC vote to designate a block of spectrum for the next generation wireless broadband that will be 100 times faster than current wireless connections? (5G)A USATODAY article stated “Advanced 5G networks will have a transformative effect on entire sectors of the economy, including healthcare, education, manufacturing, energy, agriculture, hospitality, transportation, among others,” FCC Chairman Adler said in a Friday statement.

  22. Pete Griffiths

    I suspect that the financial meltdown was great for NY tech. Instead of top talent being scooped up and overpaid by financial institutions they had to look around. We are now reaping the benefits of what they found.

  23. daryn

    Sounds pretty third tier to me 🙂

  24. Vivek Sharma

    Fred, it looks like Movable Ink (enterprise SaaS) and Maple (food) were the 2 company names that weren’t copied over from Matt Turck’s cited post. Would appreciate if you could correct. 100% agreed with the broader sentiment. NYC is in an exciting phase!

  25. Varun

    I’ve been grappling with this for a while. What is NYC’s tech identify – like really!? I think I landed on an answer when I finished watching VINYL on HBO (the ended alluding to the origin story of a very NYC establishment which I name at the end ;)Silicon Valley’s tech mythology, from where I sit, evolved from garage story >>> revenge of the nerds (office space) >>> celebrity rockstars >>> and now reaching post-modern, #peakhuman, singularity uber godmode unicorn fantasy land.NYC Tech has its own smattering of weirdos and characters and in some weird way – all of us in the city (seem to) have a say in how we shape the scene and that makes it crazy and fun! (From cursing uber/lyft/gett/via/arrow drivers >> illegal airbnb hosts trying to hook you up while taskrabitting /postmating/mapling >> hoverboards gliding by the shiny towers of Link NYC)I remember watching and how govworks tried to do municipal innovation in the 1990s or Josh Harris and his prophetic shenanigans with video streaming and ubiqutious media and then read about the origin stories of Foursquare in Anthony Townsend’s Smart cities book by or even Thomas Petterfy’s NASDAQ robot in the late 1980s as the origin story for High Frequency Trading (Interactive Brokers) – all of it had a distinct ring of cyberpunk counter-culture that reminded me of the Atari origins of S.Jobs etc. that is long gone in the valley.Anyway the answer that I rested on is that NYC tech is like the CBGB of the tech scene and our garage stories take place inside cramped offices or study halls in Manhattan or warehouses and navy yards in Bklyn. Loving every minute of it!

  26. TeddyBeingTeddy

    Hopefully its gluten free and non GMO.

  27. sigmaalgebra

    So, the ZF abbreviates Zermelo–Fraenkel? Okay.One summer at Vanderbilt, I studied axiomatic set theory, essentially the Zermelo–Fraenkel ideas but with some changes due to von Neumann. Thank you NSF for a nice summer.So, right, some of the motivation was the Russell paradox from setA = { B | B not an element of B }Then A is an element of A if and only if it is not. B. Russell is still laughing at that one.In simplest terms, the problem was that the Russell construction was self-referencing, that is, in its definition mentioned itself. And, the solution in set theory was to get rid of the self-referencing part by saying that we come to mathematics with a universal set that already has all the sets we will be talking about. Then they list some axioms, and from the axioms derive the usual properties want from set theory.So, we get rid of the Russell paradox. Do other paradoxes remain? Maybe.That’s a bit too deep below the bottom of the basement of math for me to care very much. For more, likely end up with K. Gödel, P. Cohen, forcings, etc.For your question, for anything I’m talking about here, is it related to such issues in axiomatic set theory? The answer is no!

  28. sigmaalgebra

    Maybe I could think about that if I knew what “second-order axioms” were, but I don’t.

  29. sigmaalgebra

    Get out the long underwear, heavy hats, coats, gloves, tire chains, and snow blowers — AI winter is on the way again!We expected something else?

  30. realposter

    Is the failure rate higher than in the Bay Area or Boston… I haven’t seen any numbers showing that they are different. All are pretty close – so in that case it shouldn’t be surprising.

  31. sigmaalgebra

    Suggestions:After a good course in college freshman and sophomore calculus:(1) Abstract Algebra. If you are in a college with a good math department, then might take a one semester of abstract algebra. So will learn about groups, rings, fields, the positive integers, the rest of the integers, the Euclidean greatest common divisor algorithm, the fundamental theorem of arithmetic (factoring integers), the fundamental theorem of algebra (roots of polynomials), more about the real numbers and complex numbers, finite fields, Galois theory (more on roots of polynomials), and finite dimensional vector spaces. And the course may cover some of the set theory you are interested in.If you are interested mostly applied math, then you could skip such a course.(2) Linear algebra — a lot more on finite dimensional vector spaces, matrix notation, systems of linear equations, Gauss elimination, linear transformations, eigenvalues/vectors, the polar decomposition, and more. Can do a lot more — numerical linear algebra, linear programming, least cost flows on a network where the arcs have maximum flow rates, group representation theory, algebraic coding theory, and more.There are several good authors, some on the Internet for free. Eventually, though, should go to the most respected source, P. Halmos, Finite Dimensional Vector Spaces. He wrote this when he was an assistant to von Neumann. The book is really a finite dimensional introduction to von Neumann’s Hilbert space theory.(3) Advanced calculus. The standard is W. Rudin, Principles of Mathematical Analysis, but to be better prepared for various applications you should also draw from other advanced calculus sources. But Rudin is a Fourier expert, and his book has a nice treatment of Fourier series. More applied treatments are also worthwhile and a lot of fun after seeing the solid stuff from Rudin.(4) Ordinary differential equations. My favorite author there is Coddington — he’s a world expert and a terrific writer.(5) Measure theory. The most beautiful treatment is H. Royden, Real Analysis. But after Royden, just take a second pass from the first half of W. Rudin, Real and Complex Analysis. There are several other good sources, e.g., the Soviet Kolmogorov and Fomin. Rudin has a gorgeous treatment of the Fourier transform. Measure theory cleans up a lot of rough edges left over from college calculus. Also, measure theory is also the standard prerequisite for most of advanced work in probability, statistics, and stochastic processes. Also get nicely into functional analysis where learn much more on why vector spaces are so important.Now you will be in a relatively solid position for a lot more in applied math. You will be nicely ahead of nearly all computer science profs — in my view, that math is more important for the future of computing than anything computer science has so far.go for linear algebra. So, that is finite dimensional vector spaces. Usually are using the real or complex numbers, but

  32. sigmaalgebra

    When you use freshman calculus to find, say, the area of a circle, the volume of a sphere, the surface area of a sphere, etc., really you are just using calculus to define those since we really don’t have another definition except just from intuition where we believe, say, that a sphere has a surface area.So, calculus is a way to define areas. Well, measure theory is also and is more powerful, e.g., fails on fewer pathological cases.Calculus defines areas via the integral, and measure theory is a different way to the same thing — is another integral.On nearly every function see in everyday life, the calculus integral and the measure theory integral give the same numerical values. So, in a sense, the main difference is for pathological functions.So, from calculus, we have some real valued function f of real numbers x. So we get to write the function values as f(x) and then integrate f(x).Okay, Rudin’s, Principles gives the usual sufficient conditions for the integral to exist. Want the function f to be continuous. For real numbers a and b with a < b, want to integrate over interval [a,b]. Then that interval is *closed and bounded* which means it’s *compact* which means that any continuous function on that interval is also *uniformly* continuous which means as we take finer partitions in the calculus (Riemann) integral we get convergence and the integral exists. Whew! That’s the main point of the first several chapters of Rudin’s Principles.So, with the Riemann integral, we partition the domain of the function f, that is, the X axis.Hmm, is that the only way? Nope! Instead could partition the Y-axis, that is, the values offunction f. That way, don’t have to worry about finite intervals like [a,b], continuity, compactness, or uniform continuity. Instead, darned near anything has a nicely defined integral.And that way, since are not partitioning the domain of function f, that domain can have less structure than, say, the real line. Instead, the domain of f can have so few assumptions is is just a *measure space*, that is, a set with a concept of area or *measure*. In particular, it can be a probability space in which case the function f becomes a fantastic definition of a random variable.Then lots of classic results get easier to prove and more powerful, e.g., differentiation under the integral sign, integration over the whole real line R or for positive integer n all of finite dimensional R^n, and interchange of order of integration. And, now have a nicer foundation for Fourier series and a much nicer foundation for the Fourier integral.There is a totally cute method of proof called a *monotone class* argument — three steps and have the theorem proved!So, measure theory is a replacement for the Riemann integral and is better in that it needs fewer assumptions and gets more powerful theorems. It’s also the foundation of probability, statistics, and stochastic processes. In particular, the *measure*, that is, the concept of area, becomes probability.That’s the beginning. Once get past the first derivations, there are astounding wonders to see.Measure theory was worked out by E. Borel student H. Lebesgue near 1900. The application to probability was by A. Kolmogorov in 1933. The application to stochastic processes was by J. Doob in the 1940s.Measure theory gives some nice *closure* properties and, thus, a foundation for Hilbert space — a vector space that is an inner product space where the space is *complete* in the norm from the inner product. So, intuitively, *complete* means if a sequence appears to converge, then it really does because there is a unique thing in the space for the sequence toconverge to.With just the Riemann integral, can’t have such a Hilbert space. That fact may confuse some people in quantum mechanics — the wavefunctions are supposed to be in a Hilbert space, but for that need the Lebesgue integral since some of the wave functions won’t have a Riemann integral.