5 ms·

I haven't access to the Fitch article, and I'm no specialist of the automotive market at large. I know even much less about oil. However, here are a couple of

by fab13n 10y ago

I haven't access to the Fitch article, and I'm no specialist of the automotive market at large. I know even much less about oil. However, here are a couple of guesses about how things are likely to go wrong for car makers. In a vertical market like cars, you've got many actors, from iron ore mining to the "going from A to B reliably" service, going through part makers (who actually make most of the car), car maker (brands), repairmen, dealers, etc. Very often in those markets, one actor controls a key level, protected by entry costs, secrets, regulations, exclusive customer relationship, whatever. By leveraging this stronghold, this actor extracts most of the margins, leaving just enough to other levels to let them survive. You want all other levels in your vertical market to be commoditised: Microsoft wants PC hardware commoditised, Über wants driving to be commoditised (through underpaid drivers first, unpaid robots ASAP), etc. It seems that for cars, mastering the engine is the way to control the whole vertical market. Moreover, ever-more-stringent emission + fuel efficiency regulations keep it hard to make engines, despite technical progress. Electrical cars change that, because it's (comparatively) very easy to make an electrical engine. Another asset held by carmaker is the brand and consumer relationship, and this one's probably doomed as well. Cars stay parked for 95%+ of their lives, the only reason you buy one (and it's your most expensive possession after your house) is because you want it to be reliably available whenever you want. Once cars drive themselves, it stops making economical sense to own one: for a fraction of the price, you'll rather subscribe to a fleet of autonomous cars which reliably deliver one on your front door 10 minutes after you summoned it with your smartphone. That not only means fewer cars driving a higher percentage of time: it makes obsolete the car-makers' genuine expertise in making individuals want to buy their models. I'd bet that selling hundreds of thousands of units to Kalanick is significantly tougher than making some hillbilly lust for a Canyonero. It looks like mastering level-4 autonomy will be the new way to lock that vertical market, and current car-makers don't look better positioned than the Silicon Valley to secure that stronghold. Cue the famous last words by the then Palm CEO, saying about the iPhone “PC guys are not going to just figure this out. They’re not going to just walk in”. Yes, in radical changes, it typically is an incumbent who figures it out.

angstrom 10y ago
It also does 2 other things as we've seen with other automation. The maintenance schedule can be forecast with higher accuracy (humans really make this difficult). Second, it lowers the insurance premiums as a result of higher predictability (humans REALLY screw this one up). As a comparison to existing solutions there's no room for competition if your company can't cross the moat.
jacques_chester 10y ago
> for a fraction of the price, you'll rather subscribe to a fleet of autonomous cars which reliably deliver one on your front door 10 minutes after you summoned it with your smartphone Interesting, I'd picked a different idea as the market shape ("AirBNB for self-driving cars"). Yours probably works better for a well-capitalised firm. Especially since it encourages lockin through mere inertia. There's no reason the manufacturers can't take this approach themselves. With a self-driving fleet, they can bypass their dealer channels and undercut everyone else, especially Uber, and Lyft (less Tesla). In short, it's the GE model for aircraft engine. Selling engine hours, not engines. You can charge a premium over the actual financial costs because it's simple. Plus most buyers prefer a fixed, regular payment over paying-per-use, even if doing so is financially irrational. Loss aversion and whatnot.
- fab13n 10y ago
  > "AirBNB for self-driving cars" AirBnB works great when you exceptionally need accommodation, far from your normal environment. Most people would hate to need booking an AirBnB every night of the year, in order not to sleep under a bridge. I think people can regularly move around in impersonal vehicles (as shown by public transportations), but they'd hate depending daily on borrowing a car that's clearly someone else's. They'd probably also have a problem leasing their own car when there's a slight risk that they might need it. So really, the dynamics that makes AirBnB possible doesn't seem to translate well to daily transportation. > There's no reason the manufacturers can't take this approach themselves. I'd turn it the other way: it's likely that the fleet operators will also build and repair the vehicles. But I wouldn't bet that legacy car manufacturers will be the ones cornering that market. Very unluckily for them, electric power and autonomous driving look like they'll become mainstream simultaneously. As explained above, I believe the former will breach their historical moat, and nothing indicates that they'll be the first to rock at the latter, as well as the new kind of consumer relationship that comes with fleet operating for consumers (most of them already operate corporate fleets). Anyway, unless you're a shareholder, it doesn't matter much whether the behemoth of tomorrow is called Ford, Tesla or Über. Many of the individual engineers will be the same either way, hired by the winner. I don't believe in Apple winning this: they're good at making you lust for ownership of tangible goods, but offering collective services in impersonal vehicles doesn't feel like it's in their DNA. As for Google's moonshots, I can't remember any of them being successful. Moreover, they'll probably be beaten at their usual strategy of accumulating more data than others and parsing it better: Tesla's already accumulating millions of analysable miles with their actual fleet of actual cars doing actual travels, while Google toys around SF with replicas of Noddy's car. And Musk's been hinting at interest for building a company-owned fleet (although he would be a fool to say it out loud: Tesla's all about making electric cars sexy to own). Über is also very well positioned for smart data accumulation. And IIRC, Kalanick has always been open about his plan of cornering the autonomous cars fleet market before autonomous cars exist, and hiring drivers as a temporary stop-gap (which he also openly treats like shit). > Plus most buyers prefer a fixed, regular payment over paying-per-use They currently pay a mix of the two (they pay for gas and maintenance per mile). With electricity being cheaper than oil, it's possible for fleet operators to offer an even better per-month / per-mile ratio. And I agree with you that they'll probably do just that. Unless of course, a monopoly is secured so fast that there's no point in making fleet customers happier than those of cellphone or cable operators.

by fab13n 10y ago

angstrom 10y ago

It also does 2 other things as we've seen with other automation. The maintenance schedule can be forecast with higher accuracy (humans really make this difficult). Second, it lowers the insurance premiums as a result of higher predictability (humans REALLY screw this one up). As a comparison to existing solutions there's no room for competition if your company can't cross the moat.

jacques_chester 10y ago

> for a fraction of the price, you'll rather subscribe to a fleet of autonomous cars which reliably deliver one on your front door 10 minutes after you summoned it with your smartphone Interesting, I'd picked a different idea as the market shape ("AirBNB for self-driving cars"). Yours probably works better for a well-capitalised firm. Especially since it encourages lockin through mere inertia. There's no reason the manufacturers can't take this approach themselves. With a self-driving fleet, they can bypass their dealer channels and undercut everyone else, especially Uber, and Lyft (less Tesla). In short, it's the GE model for aircraft engine. Selling engine hours, not engines. You can charge a premium over the actual financial costs because it's simple. Plus most buyers prefer a fixed, regular payment over paying-per-use, even if doing so is financially irrational. Loss aversion and whatnot.

fab13n 10y ago
> "AirBNB for self-driving cars" AirBnB works great when you exceptionally need accommodation, far from your normal environment. Most people would hate to need booking an AirBnB every night of the year, in order not to sleep under a bridge. I think people can regularly move around in impersonal vehicles (as shown by public transportations), but they'd hate depending daily on borrowing a car that's clearly someone else's. They'd probably also have a problem leasing their own car when there's a slight risk that they might need it. So really, the dynamics that makes AirBnB possible doesn't seem to translate well to daily transportation. > There's no reason the manufacturers can't take this approach themselves. I'd turn it the other way: it's likely that the fleet operators will also build and repair the vehicles. But I wouldn't bet that legacy car manufacturers will be the ones cornering that market. Very unluckily for them, electric power and autonomous driving look like they'll become mainstream simultaneously. As explained above, I believe the former will breach their historical moat, and nothing indicates that they'll be the first to rock at the latter, as well as the new kind of consumer relationship that comes with fleet operating for consumers (most of them already operate corporate fleets). Anyway, unless you're a shareholder, it doesn't matter much whether the behemoth of tomorrow is called Ford, Tesla or Über. Many of the individual engineers will be the same either way, hired by the winner. I don't believe in Apple winning this: they're good at making you lust for ownership of tangible goods, but offering collective services in impersonal vehicles doesn't feel like it's in their DNA. As for Google's moonshots, I can't remember any of them being successful. Moreover, they'll probably be beaten at their usual strategy of accumulating more data than others and parsing it better: Tesla's already accumulating millions of analysable miles with their actual fleet of actual cars doing actual travels, while Google toys around SF with replicas of Noddy's car. And Musk's been hinting at interest for building a company-owned fleet (although he would be a fool to say it out loud: Tesla's all about making electric cars sexy to own). Über is also very well positioned for smart data accumulation. And IIRC, Kalanick has always been open about his plan of cornering the autonomous cars fleet market before autonomous cars exist, and hiring drivers as a temporary stop-gap (which he also openly treats like shit). > Plus most buyers prefer a fixed, regular payment over paying-per-use They currently pay a mix of the two (they pay for gas and maintenance per mile). With electricity being cheaper than oil, it's possible for fleet operators to offer an even better per-month / per-mile ratio. And I agree with you that they'll probably do just that. Unless of course, a monopoly is secured so fast that there's no point in making fleet customers happier than those of cellphone or cable operators.