The headstart that both Tesla and Google have is test data. Tesla has more, because they've collected data from production cars from the get go. Google likely h

The headstart that both Tesla and Google have is test data. Tesla has more, because they've collected data from production cars from the get go. Google likely has more mature software (pure speculation), and it has comprehensive mapping data, but it doesn't have 100k customers providing them with useful real-world information. It does, however, have millions of Android users, but that's not the same. I believe all new Tesla vehicles send back autopilot data to the mothership. There's some interesting comparisons. On the one hand you have Google, which is primarily a software company with hardware interests. Then you have GM which is an auto company with software interests. And you have Tesla which positioned itself in the middle, aiming to both build cars and software. As a result, Google has a lot of core software expertise and infrastructure (particularly for machine learning). GM has the production capacity and the ability to roll out features to potentially millions of new vehicles. Google had to buy in automotive people. GM had to buy in software people. Tesla had to buy in both, but they did it from the very beginning with clear goals in mind. From an outsiders' perspective, building a car that can drive by itself isn't terribly difficult nowadays. If you have the cash, you can buy a team that will build it. The challenge is getting from that stage to a car that can reliably drive better than a human in all driving conditions and deal with unexpected events. That will take much more time, and much more data. There are other aspects like getting the sensors down to a price that people can actually afford.