This was always a clear threat. While Uber is also working in this space, it always seemed very clear to me that once you “solve” driverless cars, creating a network similar to Uber is relatively trivial, assuming human capital is not required 1:1 as it is with Uber.
In theory, to the customer, it doesn’t matter if a human or machine is driving, so you’re no longer dealing with a two-sided network, making adoption substantially easier, especially for a company like Google who can deploy massive capital.
The thing I never understood is how Uber’s investors rationalized this and thought it wouldn’t happen. Did they think self driving cars would not be a reality? Did they think Uber could get there first? Did they think there is still a profitable enough gap between current situation and the dreiverless future to hedge in case Uber doesn’t get there? Is the Uber brand and tech worth so much that they’ll get something back as an acquisition for whoever does get there first?
Seems like exactly what they thought. Remember Uber acquiring Otto and Levandowski? AFAIK Travis was pushing hard for autonomous driving but accidents like the fatality in AZ and missteps like Levandowski's fraught relationship with Waymo led to deemphasizing autonomous driving for the time being. Dara's main job seems to be to take the company public, which requires scaling and fixing the margins on the products they already have in place.
I don't work at Uber and perceptions are my own, so curious if any Uber employees agree.
At the current cost of a self-driving car (~$250K), it would be ridiculously expensive to build a fully self-driving ride-sharing fleet [1]. So expensive that I doubt it's possible to make real money that way right now.
Self-driving cars are a long game until companies can bring down the costs. It might be years before that happens. Uber has the advantage of being able to ramp up self-driving cars as part of their existing fleet until then.
I believe that the cost to produce a self-driving car will be coming down, but let’s look at the $250k price tag.
Over 6 years, that comes to $114/day. 1 ride per hour at $5 per ride would hit $120/day.
Assuming a 2 mile ride, that would be 105k miles which is well within the car’s usable life. At 3 miles, it would be 160k miles which is still within a car’s lifespan. It looks like Uber is $1.35/mile with a $2.10 base fare and $1.85 fee. A 3 mile ride should be able to get $5.
Plus, it’s really about the long run. Operating margin might be negative for a bit, but the cost of the technology and manufacturing will come down significantly. Still, even today, I think $250k just isn’t that bad. I think most drivers will do a lot more than $5/hour in gross revenue.
It would be really expensive to make a fleet, but the economics are so compelling, even at high prices. I mean, Americans are often spending $35,000 on a car. If a self-driving vehicle can service the needs of 7 people, it can be cheaper than car ownership.
While the price of driverless tech might be high now, the prices of car ownership and human labor are both very high as well and only one of those three prices is likely to decrease over time.
Driverless cars are "always on" so it's not an apt comparison because you're going to always get more from less. Besides, ridesharing is already subsidized right now, so bottom line isn't an issue in the market.
Likewise, ride-sharing with self driving cars is even easier since if you own a self driving car it can work on it's own while you are not using it, which is most of the time. There's no reason a similar asset sharing model won't spring up after economies of scale.
You could also argue that it's cheaper to pay workers with their own textile equipment, rather than spend the CapEx on a weaving loom, yet here we are.
It's not trivial when we don't even know what sort of regulation will be required to maintain these vehicles' road legal status, let alone banking on future economies of scale. Would it be a simple license fee? Perhaps an inspection? An age requirement on the vehicle? Or perhaps the worst (and very possible) outcome, mandating that an undistracted driver be sitting behind the wheel at all times, ready to take control.
It would take one trivial piece of legislation to turn this technology into a souped up cruise control rather than the world changing technology that its backers insist that it is. How could you even fight that sort of regulation without your argument boiling down to 'we promise we don't need human oversight.' One lazy 'think of the children!' retort later, and it's banned faster than mango juul pods.
I don't understand the investor confidence here. To me this seems like basic research, critical for future technological developments, but a cash sink with no guarantee for profitability. Is this just a rat race between the giants throwing cash at this?
The likelihood of such a legislation on a national level is unlikely. Instead, you will have cities that refuse to adopt self-driving, maybe because they were burned with a tragedy (with no regard to statistics), maybe to protect human jobs.
Large companies can adopt in other cities, or if need be, other countries eager to change their quality of lives.
If this is trivial to you, what do you consider moderately difficult? What's sufficiently difficult that you have to spend a single day thinking about it? How about an open problem requiring careful planning and months of effort?
It's like the word "trivial" has no meaning anymore. All I can tell from people using it these days is that they're very confident in what they're about to say.
What I consider non-trivial is tangental to comparing the cost/benefits between autonomous driving and ride share services. The non-trivial part of this is [nearly] solved, as is evident by this Waymo soft launch.
The self-driving cars may be available to give rides 24/7, but if 90% of the demand is 0800-0900 and 1600-0200, they may end up sitting idle for twelve hours each day anyway.
You probably need over 100x the vehicles at 5pm as you do at 4am. You have to find a balance of how much CapEx is worthwhile -- probably more than you need to cover the 4am shift, but far less than needed for the 5pm shift.
Factor in the cost of a driver, and the $250K price tag will look super cheap if not sensible.
Right now, rideshare companies keep ~40% of the ride revenue (not including vehicle payments). Which means, 60% goes to the driver. Now if you consider operating costs (fuel/gas/insurance/tolls/oil change/maintenance) at 30% (of the total revenue), there's still the remaining 30% that the driver takes home. Even if you think the driver is working 80 hrs a week, that's not even 50% of the total number of hours an autonomous car can work (24x7 = 168 hrs a week).
With a fully autonomous car, that remaining 30% and lower maintenance costs combined with lower insurance costs will mean probably close to 40% of the current ride revenue will be saved. If you operate the car for 160 hrs a week (8 hrs for fuel/maintenance) then it'll probably be ~60% of ride revenue (at current rideshare rates) that'll be pure profit.
As cost to produce these cars (basically it's the separate technology package added to regular cars) goes down, they can continue to keep the service super affordable while recovering the upfront investment in the $250K car in a few years. Instead of having 10 Chrysler vans on car dealer lots/storage, have one $250K car in service a day after it is manufactured. After that maintenance costs may increase slightly, but the rest is pure profit, which creates a positive feedback loop to
reduce vehicle price --> reduce final fare paid by passenger --> attract more users --> more vehicles needed --> Leads to larger scale production and reduction in price--> repeat.
At some point it'll be cheaper than public transit and then owning and driving a car becomes a need only in regions / places / conditions where autonomous cars wont work. Until then, non-autonomous cars ensure good margin for autonomous cars to thrive.
FWIW, I believe Uber has zero advantage with it's existing fleet. All they have is a large number of indentured drivers (due to subprime loans given to them for their cars). In fact it is a disadvantage for Uber, because as fares for ride share goes down, it starts making lesser sense to operate those cars. they'll just stop making payments and return / trash the vehicle.
I don’t think you need same number of self driving cars as drivers and the ratio might be way higher than 1:3. Plus I think you have a math error and the total cost in your example is $250B which is doable for Google.
>So expensive that I doubt it's possible to make real money that way right now.
I don't think anyone is interested in "mak[ing] real money that way right now", they want to get a foothold in, or capture a swath of, a market which definitely will make them money in the future. A tremendous amount if they play it right, even if it means operating at a loss for the next few years.
I believe you're referring to an issue that Teslas have experienced? Waymo cars are way more sophisticated, both in software and hardware. They use LIDAR to see large static obstacles and don't ignore them.
We now are going to get Uber complaining that Google is taking their jobs, and not being regulated like they are, and how technology shouldn't put people in the streets.
Humans drivers could be necessary to... just lend the car. Not for their driving skills but for the money they invest. It’s much easier to let car owners take the risk of investing in a car, maintaining it, repairing it (for the moment they don’t repair themselves), cleaning up the Saturday night mishaps. It could be hundreds of billion dollars to populate the world with Google-owned self driving cars.
> Humans drivers could be necessary to... just lend the car.
Unless the humans are losing money by doing so (which may be th case, but probably doesn't make a sustainable business model), the car service is ultimately fully paying for the car for the period it uses it, so it doesn't lose anything by doing so up front. It increases the up front capital requirements, but Google has more cash on hand than the market cap of some Fortune 500 firms, so that's not really an enormous issue for Google.
> It’s much easier to let car owners take the risk of investing in a car, maintaining it, repairing it (for the moment they don’t repair themselves), cleaning up the Saturday night mishaps.
Well, except humans lie about the last bit to get extra reimbursement, whichnpisses off customers. That kind of thing and other driver frauds is a compelling reason to get independent driver-owners out of the loop.
Consider the fact that car manufacturers don't generally operate taxi services. For the same reason that Toyota sells Priuses (Prii?) to taxi companies, Waymo may choose to sell its technology to car manufacturers and via them to taxi services.
In theory, to the customer, it doesn’t matter if a human or machine is driving, so you’re no longer dealing with a two-sided network, making adoption substantially easier, especially for a company like Google who can deploy massive capital.
The thing I never understood is how Uber’s investors rationalized this and thought it wouldn’t happen. Did they think self driving cars would not be a reality? Did they think Uber could get there first? Did they think there is still a profitable enough gap between current situation and the dreiverless future to hedge in case Uber doesn’t get there? Is the Uber brand and tech worth so much that they’ll get something back as an acquisition for whoever does get there first?