It's because the car industry is about to experience two fundamental revolutions simultaneously. The move to electric drivetrains which will completely change the technical architecture of the hardware, and the move to self-driving. The incumbents have no built-in advantage in either of these areas, which makes it ripe for disruption on two different axes at the same time.

If a move to a new manufacturing process such as carbon fibre gets thrown into the mix, that's three simultaneous revolutions.

> More and more people live in an urban area and more and more young people use public transportation way more often, than cars. Additionally carsharing makes owning a car in a german city kinda obsolete.

It is this. What you're missing is that in the whole wide US, the number of European-density cities where most people can choose to forgo routine use of a car can be counted on one's fingers. The common layout here is a few walkable commercial blocks "downtown", that generally serve as single-destination stops for drivers who live further out.

Self driving is what's required to make the car sharing frictionless and on-demand in less-dense areas. Currently this role is being done with human owner/operators; replacing the humans with software will reduce prices and consolidate ownership of capital equipment. (Interestingly enough, it will also decrease the stickiness of supply. Instead of needing to slowly recruit driver mindshare for a new intermediary, a straightforward investment in a fleet of cars suffices)

That being said, I think this article is a bit hokey. Apple "pivot"? mm hmm.

There is a contrarian argument that self-driving cars will actually INCREASE the demand for vehicles. If a vehicle is self-driving, it can not only act as a taxi, but also act as cargo transporter, delivery guy, etc.

It's not really that much of a contrarian argument.

Basically, here's the deal:

Right now, the cost to own a car is CO and the value received from owning a car is VO.

The cost to rent cars from a taxi-like service is CR and the value received from renting cars in a taxi-like service is VR.

For most people, CO < CR. That is, it's usually much cheaper to own than to rent. With perhaps some exceptions if storing your car is super difficult in your locality.

The VO and VR relationship is much more complex. There are advantages of each model (you can take an Uber when you're drunk, you don't have to look for parking, but on the other hand you can't leave your crap in it and you have to wait for it to arrive and you can't take your kid in it).

Based on the CO < CR and VO <?> VR distribution across people, we have a certain balance where taxi-like services are pretty big, but ownership is the dominant model.

In the autonomous world, VO > VR almost all the time. That is, you get all the advantages of current ownership and pretty much all the advantages of current rental. But the cost of ownership presumably increases (you have to pay for a sensor package now) and the cost of renting presumably decreases (since you no longer have to pay for a driver). So the question is how much do those costs end up changing? Is the sensor package expensive or cheap? How many human costs can the Ubers of the autonomous world actually cut out, and how much utilization can they get out of each car? If the CO increase is small enough or the VO decrease is small enough, you expect to see more car ownership. If not, more car rental.

> public transportation

because private companies don't generally dabble in public infrastructure :)

I think at this point it's less about cars or electric cars, but more about self-driving cars(that happen to be electric). In the U.S. at least, we have a MASSIVE roadway infrastructure that's not going anywhere for a long time. Utilizing that probably looks like a much easier path to automated transportation than does building up the rather sparse network of railways.

Imagine this future in Europe: pretty much no one owns a personal car. When you want to go somewhere you order a car on your phone and within 10 minutes a driverless car arrives and takes you to your detination. Your music plays through the sound system. The back seat is comfy and you have plenty of room to work or surf the internet. The trip could be to the hardware store, or hundreds of miles to visit family.

I feel like OP understands this - the questions is that why do any of that when the younger generation is rapidly urbanizing, in which case "order a car on your phone, wait 10 minutes" seems silly compared to "walk out front door, onto train, be at destination even sooner".

The self-driving utopia is primarily a suburban-centric dream, in many European cities (and a small handful of American ones) it's likely to be less efficient than what already exists.

Around here for example, the subway runs every 2-3 minutes and moves faster than cars - perhaps even self-driving ones. Trips are generally short enough that being able to work/surf the internet is a pretty marginal benefit, and the hardware store is so close that it would be nonsensical to use any vehicle to go there. All of the above pre-supposed long travel times and having to take a vehicle for even mundane every tasks - this is untrue for many cities, where cars (self-driving or otherwise) will likely be strictly inferior to the utility offered by walking or mass transit.

Where I live in the UK, I guess it is part rural and par urban (it's a bit of a sprawl.) Everyone owns cars. Buses run, but your journey is going to take a very long time. Last time I checked to get to the nearest city took over an hour on the bus and 15 minutes by car.

City subways systems take decades to build. Public transport is a shambles outside high density cities, and in many of those. Here in SF, it takes me an hour to get downtown by bus or subway, and 10-15 minutes by bus.

Right, but the idea is that there seems to be a large preference towards high density cities going forward, so it seems like self-driving cars improve a lifestyle that is to some degree actively being abandoned.

I think OP's point is (to paraphrase) "why all this when the population is moving into high density cities where this complex infrastructure already exists and presents a superior choice to cars, self-driving or otherwise".

Or from a more global perspective "why all this when the US is the only major market that is likely to continue having a mass suburban population in the long term, and the rest of the world has already urbanized and will continue to do so, already has highly robust mass transit infrastructure, is actively building even more of said infrastructure, and that is rapidly attracting even more population?"

There is a particularly common dream among American urbanists that suburban America will somehow rapidly urbanize - I don't think that's particularly true, but it's certainly true the rest of the world has urbanized, and will continue urbanizing rapidly. So the question is if self-driving personal cars (as opposed to say, commercial cargo vehicles) will largely be an American phenomenon, and if so, doesn't that present fairly modest total addressable market for a company the scale of Apple.

And I'm living in a high density city, where the public transportation is barely usable.

EDIT: that should be 10-15 minutes by car.

The future seems likely to involve a fleet of autonomous vehicles that we use on-demand -- why not own and operate the fleet? Not to mention owning the technology powering the autonomous cars.