What makes the perfect AI startup? This week we look into the AI startup craze and strategize how to launch an ideal AI startup.
What do Siri, Alexa and AlphaGo all have in common?
They all stem from startups. Apple bought Siri in 2010, Amazon bought the technology that underpins Alexa in 2012 and Google bought Deepmind, creators of AlphaGo in 2016.
When you dive into how much of the supermajors AI is from startup related acquisitions, it is significant. Amazon’s image recognition API, Rekognition, is literally just a company that Amazon bought. Facebook recently introduced natural language processing into Pages, which allows AI to answer questions your customers may have and was also literally an acquisition. Then there are the acqui-hires that we have talked about, where companies paid millions of dollars just to buy talent.
It’s no surprise then, according to CB insights, AI acquisitions were up 44% from 2016 to 2017 and up 422% since 2013.
Venture capital fund sizes are also getting big. Lee Kai-Fu, an ex-Googler, recently raised $500mn for an AI based VC fund in China and Andrew Ng, co-founder of Coursera, also recently raised almost $200mn just for AI startups. As the amount of capital in the sector increases, the number of AI startups is also exploding. Angel.co lists almost 8,000 companies with some sort of AI affiliation, with almost every industry covered. Just in autonomous cars, Angel.co put together a list of the top 30 companies “To Watch.”
Let’s launch an AI startup!!!!
One of the biggest problems AI startups face is not only are they going to have to compete with each other, but they are going to have to compete with the supermajors as well. After Google’s developer conference this week, TechCrunch mentioned Google was going “All-In” on AI. Good luck competing with that.
I don’t care, I still want to launch an AI startup
If you are still undeterred by this and want to create your own AI startup, what would be your best bet? Right off the bat, you would probably want to be in an industry where the supermajors can’t just clone your product. If you are trying to create a new product that responds to email, just remember Google has everyone’s emails. But, what about assessing an industry that the supermajors don’t play in. It’s no surprise that when looking at the top AI startups with the most money raised so far, three of the top 10 are in Fintech, as the supermajors don’t do much business there.
Another smart move would be to pick a niche where you can apply domain specific knowledge. If you used to work in the airline industry and want to apply AI there, what about developing some sort of AI that helps traffic control carriers better do their job. This way you understand how all the data comes together and what pain points/business challenges AI can improve.
Or, just move to another country
One of the interesting things about the most valuable AI startups in China is they are all doing generic image recognition, something that can be easily cloned by the supermajors. However, these companies have an advantage as they are most likely tapping into extensive government databases of images to train their algorithms. As long as you can convince a government to give you funding and access to data no one else has, you’ll be just fine.
They party keeps going
With tech companies posting quarter after quarter of blowout earnings and billions of offshore money coming back to the US, expect the AI binge to continue. But just give it some time to think about how you are going to be different than the other 8,000 companies.