One of the reasons Uber is "losing" money is because they've made a huge bet on autonomous vehicles, buying the entire CMU robotics department, and Stanford, the winner of the 2005 Darpa Grand Challenge isn't far from Uber's headquarters, either.

30 years ago was 1986; I don't know about you, but my Internet usage looked a lot different back then, and computing power was drastically different. Fully autonomous vehicles may still be a ways off, but imperfect partially autonomous is already here, with Tesla leading the the way for high-end consumers.

It's going to work fine in confined spaces, i.e. loading/unloading cargo, transporting goods inside industrial parks, even pre-defined drone paths for item delivery with collision avoidance etc. And the more self-driving cars there are, the more could they be globally orchestrated with phasing out human factors. Cars are now full of sensors and increasingly communicate with each other. Imagine they would notify each other about their trajectories, state, etc. when they get closer/pass by.

Most of the luxury car makers have Level 2 systems shipping now. Google and Volvo have good Level 3 systems in test. Volvo plans to deploy 100 Level 3 cars next year. Google and Chrysler are building 100 Level 3 minivans in Michigan. Level 3 isn't that far away.

While true, Level 3 will be a (very) nice capability once available. And I agree it's relatively close. But you're probably not enabling new businesses (for general road use) until you have full door-to-door autonomy. Which was what the original comment was talking about.