Why startups are going to win the self-driving car race…

I was pretty excited to get a Tesla Model X a few months ago. Mostly my excitement was about the auto-drive feature. After six months, I would say that excitement was completely justified. I think a Tesla with auto-drive is the best commute vehicle ever made. Heavy freeway traffic is where it really shines. You aren’t going fast enough for any mistakes it makes to cause serious damage and it does the hard part and saves you a ton of mental energy.

That said, I’ve thought a lot about being an early adopter when the risk is my safety and the safety of my family. My wife is much less of an early adopter and is constantly worried when we are on auto-drive. For her, something that hasn’t been true before (cars that drive for you), is unlikely to be true now. I think most people will treat self-driving this way, putting themselves squarely in the late adopter category over safety concerns.

Ironically, that slow adoption gives startups like Uber and Tesla an enormous advantage over large companies like Google, Apple, and GM. There are four areas where I think that advantage plays out:

1 — The safety risk is much worse for the big companies, because their brands are built around stability and consistency, and having any safety problem would hurt them a lot. Tesla and Uber customers are early adopters, who are making a conscious decision to opt for the convenience of self-driving even with the safety risk.

2 — The market will be small for a long time relative to the overall car market. For huge companies with huge revenues, focusing on something for a decade that won’t affect their quarterly reporting is incredibly difficult. For Tesla and Uber, it will have an enormous impact on their revenues and profits, so they will have intense focus.

3 — Given both of the above, we get to the checkmate in this game. Data (and smart people that know how to use it) is the main game in self-driving. Getting to the 1000’s of edge cases in real-world navigation is going to be based on providing your learning algorithms with enough data on those edge cases that they can adapt to them. Unless a company is all in like Tesla with auto-drive or Uber in Pittsburgh, they won’t collect the data they need to learn fast enough.

4 — The moat created by being first to collect data and improve your driving algorithms will be very deep and wide. Safety is the key thing. How many people are going to be willing to buy a Google-powered car which is 5% less safe than a Tesla or take a GM/Lyft which is 5% less safe than an Uber? Not me.