AI Hub


What makes an AI hub? This week we explore tech hubs, which cities dominate in AI, and give tips on how to make your city into an AI Hub. 
I wanna wake up, in a city, that never sleeps
This Cloudy has undergone many revisions; hours worth of revisions, actually. Originally I wanted to write about how New York is a tech hub and ask the question of is New York an AI hub? The inspiration came when towards the end of November, the talk of the NYC town was Amazon’s decision to move “part” of its HQ2 to NYC. Plenty of articles arose about how the decision made NYC into a tech hub, while others wrote about how NYC is already a tech hub: NYC is second to Silicon Valley/San Fran in tech patents and startup funding and while New York lacks a headquarters of major technology company, Google/Amazon are expanding into NYC at a significant pace. 
Then I started thinking, if you are a tech hub now, does that automatically make you an AI hub? The answer that I came up with, and is something that you could certainly do more research on is: sort of.
What constitutes an AI hub?
In order to be considered an AI hub, you need people working in AI/ML, companies that do this type of work and a healthy startup/university/venture capital ecosystem to introduce new AI/ML ideas; essentially all of the reasons you need to be considered a tech hub. 
In order to determine which city leads in the race to AI, I took data from LinkedIn on how many people have “machine learning” in their profiles, data from CBRE to assess number of technology workers (different sources were used for EU) and looked at data from Crunchbase on the number of AI/ML startups.

No surprise, the Bay Area leads in all of these metrics, and the highlighted green are cities which have a high percent of tech workers doing ML. 

Location Tech Workers ML Workers Percent ML Startups
SF Bay Area 329,200 80,793 25% 1,499
London 300,000 26,492 9% 612
New York 254,300 48,643 19% 452
Washington DC 248,200 17,391 7% 102
Toronto 241,400 17,064 7% 151
Paris 200,000 20,364 10% 154
Dallas 160,800 12,023 7% 26
Seattle 145,100 29,348 20% 82
Chicago 140,100 17,222 12% 73
Atlanta 134,800 11,024 8% 43

To validate this, I looked at similar research done by an interesting company called CognitionX, which looks at which cities have the most AI suppliers (AI supplier is defined as “one who sells at least one AI product whether or not they also sell non-AI products”). While their research was originally focused on London, they also find the Bay Area as having the most AI suppliers, followed by London and New York.

To answer my original question
Yes, New York is an AI hub, as it has a high percent of its total tech workers doing ML/AI, lots of ML workers, and a bunch of AI startups. But what other things can we learn from this data and how can your city become an AI hub?

First: Do as the Bay Area does
It’s no surprise the Bay Area ranks #1 as an AI hub as they are just doing what they do best: always looking for the next best thing and constantly evolving. The dominant industry in Silicon Valley has changed about every ten years, starting with defense contracts in the 1950s, then switching to integrated circuits, then personal computers, internet and finally social media (There is a nice chart at this link). You could make the argument that 2010 was the era of AI in Silicon Valley/Bay Area as numerous measures such as AI startups formed, AI/ML conference attendance and robotic installations all start their incremental growth in the 2010-2012 range. This chart says it all

Another trait of the Bay: connectedness. In a book called Regional Advantage, AnnaLee Saxenian describes how many of the Bay Area companies during the integrated circuit era, despite competing in the same industry, were extremely collaborative and would often share ideas and insights. Research done by Startup Genome backs up this claim, with the San Fran Bay area having an extremely high global/local connections.

This very much applies to AI, as many of the Bay Area companies like Google, Facebook and Uber are open sourcing their AI code. Here’s a great example: while not all AI projects, Google has made 1,391 projects open source, while Goldman Sachs has made 14

If you are not going to open up code, at least collaborate. While not Bay Area based, Amazon holds an annual internal AI conference which involves thousands of engineers, many of which a source tells Cloudy are flown in from all parts of the world. 

Second: If you want to be an AI hub, its up to the leading tech companies to make it happen. 
Who employees the most AI/ML people in leading AI hubs?

Seattle: Amazon, Microsoft
San Fran: Google, Facebook, Apple
New York: Google, IBM, Bloomberg, JP Morgan

New York is the only exception where a company not originally headquartered in a city accounts for most of the AI talent, but typically the most “innovative” companies within a city are driving the hiring. This is especially true in Seattle where Amazon and Microsoft, account for 41%, YES, 41% of the ML workforce. 

Third learning: favorable regulation certainly helps
Why does California have 117 autonomous vehicle startups and New York only has 8? California was the third state to pass autonomous vehicles and when New York finally got around to approving them five years after, the details were extremely prohibitive: each autonomous car had to be accompanied by a police escort, the law only permitted testing for a year and there was some pretty shady “we’ll approve you only if you open an office in NYC” issues going on, as Cruise Automation was required to lease office space in NYC to be approved. The original California self driving laws only required a person to be present and for you to file some paperwork, thus making it glaringly obvious why California won so many companies. 

Fourth learning: universities matter, and its especially important if they are intertwined with industry
Referencing Saxenian’s book again, she goes into great depth about how connected Stanford University is to industry, and that is confirmed by data showing Stanford has the most entrepreneurs/capital raised since 2008, with UC Berkeley coming in second. If you want to have your city grow into an AI hub, you need to go to lengths to connect what is coming out of the research lab and investors that are willing to fund these technologies. It will be very interesting to watch the brand new Cornell Technology hub built in NYC, because they are trying very hard to bridge industry/academia and NYC universities do not rank as high as entrepreneurs. 

To recap
If you are a city that wants to build an AI hub, give incentives for large companies to hire AI talent, introduce favorable AI legislation and have plenty of happy hours between researchers and venture capitalists.

Doesn’t sound so hard, right?

Facebook Comments