Vehicles that are one forth the mass can be estimated to cost approximately one forth to purchase."
Lol, nope. A lot of parts of vehicles have an inherent fabrication cost, something which wouldn't scale unfortunately.
I don't know enough about vehicle manufacturing to say how realistic that cost estimate is, and I'm guessing neither do you.
Well, I'm a mechanical engineer and almost minored in materials enginering (caveat: currently employed in oil and gas construction) and my major hobby is modifying cars. So i'd say I have bit of an idea...
But, we do have other precedents to compare to. For example:
http://en.wikipedia.org/wiki/Tata_Nano
Looks an awful lot like the podcar, doesn't it?
http://www.huffingtonpost.com/2012/10/15/tata-nano-3000-car-coming_n_1967753.html
Short version: car costs $3000. They're adding features for a planned 2015 US release that will up the price to $8000.
Hold up there, that isn't a fair comparison to make. You're comparing a car made in a developing country to cars made in first world countries, the costs associated and the end product are going to be both wildly different. I doubt that Google could get away with using cars with the fit and finish of the Nano, not cars with the safety rating.
Remember, in your example, the car's only replace half the cars in the city, safety rating is going to be just as important as it would in a normal car.
I think it's pretty reasonable to suggest that an electric robo taxi with no mechanical interface like steering wheel, pedals etc might be cheaper to produce.
Honestly i'm not sure you could save that much money. You still need to have all of the mechanical components of a car, eg steering rack, throttle body/control etc. the difference is that a computer is inputting commands not a human. you might save a few hundred bucks at most i'd suggest. The cost of mass producing sticks of metal to be pedals isn't much.
But at the same time it also has to have computers, camera, etc. We don't know yet how much that's going to cost. But we do know that the most expensive element of the system, the lidar, which cost $70,000 in the original prototypes is expected to cost $250 per unit. That's a pretty big cost reduction. And we know that Audi's system cost only a 'couple hundred dollars. And simply looking at the general trend of diminishing costs in electronics in general, I think it's reasonable to suggest that the costs don't need to be astronomical. Cameras, processors, LCDs, GPS and wifi are not expensive.
I wouldn't quite be so optimistic about the tech there. Your article mentions that "a" LIDAR unit costs $250, but doesn't indicate that a LIDAR unit of the type Google needs (eg, resolution, speed, useability in different conditions etc) is going to cost that much.
The AUDI system that costs "a few hundred dollars" uses a camera, its a different kind of system entirely. Kinda neat though, this is the sort of slow, small methodical expansion I expect to see as the true introduction of "self driving cars", not some kind of massive, instant world changing event.
So let's arbitrarily stick with the $8000 figure. Hey...let's bump it up to $10,000 per unit, because why not? 18,000 vehicles at $10,000 each? That's only $180 million.
Google could do that every single month, year after year, and it would still only be roughly half of what they've been averaging in acquisitions over the past five years.
So maybe it'll be $30K, or $40K? I'd say it'd be more than your arbitrary 10K figure for sure! don't forget you need bigger vehicles too, they increase the average.
Can Google afford to buy all that? sure. But because these small price increases add up. The costs compared to traditional transport, as per that report, go from being 20% to say, 40%, or 50%. And this means that the self driving taxi fleet potentially goes from being a profitable venture to an unprofitable venture.
Regardless of how much money Google has, as a listed company, they have a responsibility to their shareholders to make a profit. You will find that all of the acquisitions that Google has made are made because they will pay off (eg, youtube). Now, Google has poured money into self driving cars, and bought a lot of robotics companies of late. However these things combined do not mean that Google is aiming for a self driving car taxi fleet, just that they see money in making self driving cars and robots.