[TuringNYC]

However, something as epic as self-driving cars are one of those things that come once every two generations. SDCs, if achieved, have the potential to change the entire American way of life and massively disrupt society. Start-ups love to disrupt and this would be the mother of all disruptions.

So, for that reason alone, I think firms are willing to bet a sliver of their portfolio. If I was a family office I certainly would do so.

[CPLX]

Right. But it's the second half of the business strategy that I outlined above that's key to the discussion, ie the idea that paying billions of dollars for market share in this field makes sense. That's literally crazy, the network effect and lock-in at scale is really modest at best.

Sure the more cars you have the better the service can be, but it's trivially easy for a competitor to come along at any time and attack your most profitable market segment in a given city, and trivially easy for any customer to switch as easily as clicking on a different app and glancing at the estimated time and price. That's going to be true forever, this isn't a market that will ever have a defensible monopoly position.

Investing some amount to hedge in self-driving cars could conceivably be defensible, but looking at what's going on it feels like that's more of an rationalization for their present behavior.

They've been lighting money on fire subsidizing rides for a few years, and have hunted around for a plausible excuse. One is the "pool" functions, as that has a slightly more plausible network effect story, and the other is self-driving cars.

Both appear to be post-hoc rationalizations designed to provide some plausible story for why they need to borrow another couple billion dollars.

[TheOtherHobbes]

I wouldn't say it's trivial for a competitor to take over.

The Uber app is still best-in-class, and all my friends now say "Get an uber" instead of "Get a cab" because the UX is so much smoother.

Generally it's extremely difficult to dislodge an incumbent from a market slot. Newcomers can only compete on price, UX, and brand recognition. Ideally all three need to be significantly better than the incumbent to have a hope of taking over. Without all three the best a newcomer can hope for is a small slice of the pie.

So market share definitely has value in the abstract. Unfortunately in Uber's case it has negative economic value because of the costs/subsidies.

Then again I suppose it's possible Uber has always been a cunning plot to take VC money and spend it on subsidised transport. If so, it's definitely been a success - for now, at least.

[ExactoKnight]

What you are saying is right about Uber's current business. But that is not true for self-driving cars.

The infrastructure costs of deploying a fleet of cars to compete with Uber if they gain market share will be massive, all but assuring they will hold a monopoly position for years unless a decentralized competitor could actually become reliable quickly (I doubt it).

[fmihaila]

Once you have a self-driving fleet sure, you can start printing money, but it makes no sense to sacrifice revenue today by starting an unsustainable price war long before the tech is ready. All this does is shorten your runway, and you don't even know how long a runway you will need, given that reliable, fully autonomous driving is so hard.

[CPLX]

> Once you have a self-driving fleet sure, you can start printing money

Why?

Are drivers who make $10-15 an hour so ludicrously expensive that saving that money fundamentally changes the business?

Are the carrying costs and maintenance costs and depreciation costs of self-driving cars likely to be lower or higher than the cost of a 2017 Toyota Camry? How about the regulatory and insurance costs?

Is there likely to be some magic secret that allows one company to dominate self-driving cars, or will it resemble the historical markets for transportation devices, whereby there are dozens of companies who make different offerings of similar technology, and a network of component suppliers and hardware and software companies that contribute?

I guess it captures a unique place in our imaginations, but this topic has an unusually severe infestation of magical thinking for some reason.

[krakensden]

I can see an argument that a vertically integrated car manufacturer and taxi service provider would be hard to beat. Still makes Tesla and GM the companies to worry about, not Uber. The manufacturing part is much harder than the app part.

[fmihaila]

> Is there likely to be some magic secret that allows one company to dominate self-driving cars [...]?

I actually agree with you that this is not a given, but it wasn't clear from my phrasing. I should have started my comment with "Even assuming you will print money with a self-driving fleet, etc".

That said, whoever deploys a self-driving fleet first should enjoy a significant cost advantage against human-driven taxis, at least for the initial period before competitors finalize their own transition. Insurance costs should get lower if SD cars prove to be safer (if not, they wouldn't pass regulation), and taxi customers are very price sensitive, so the $10-15 per hour cost advantage will certainly matter. Taxi drivers hate the price competition from upstarts already. If anybody figures out how to make those prices economical, they win, at least until the market commoditizes itself at a lower price level.

[elif]

Given your $10-15/hr rate, the monthly insurance costs for a vehicle are paid for in less than a day. As SD cars prove themselves, they may even have lower insurance costs than humans.

Since drivers can't work 24 hours per day, you are paying depreciation on ~three 2017 Toyota Camry's, not one.

[danielweber]

Assuming self-driving cars are just about to work, why is Uber significantly better positioned to profit off of them than other people?