Hacker News new | past | comments | ask | show | jobs | submit login
Uber seeking to buy self-driving cars: source (reuters.com)
210 points by martinshen on Mar 18, 2016 | hide | past | web | favorite | 205 comments

TLDR: It's not self driving. It's data collecting cars for now.

I know a little bit about this. It's not "self-driving" car per se. You know with current technology we can't have completely driver-less cars. But to get there we need a lot of data. Uber and Lyft saw all the rides on their systems and thought if we had a lot of sensors in cars and recored all of that we could have the winning factor for self driving cars which is data.

In order to record that data they need customized cars. They also need to own the car to own the data associated with it (this is my assumption).

That's not the TLDR.. that's not in the article at all. You are correct however that it's unlikely that Uber would commit a large order for a car that doesn't exist yet.

It would make more sense if they were ordering their own fleet of standardized, sensor-laden, optimized UI cars. But that's not what this (dubious) news report is about.

I meant my comment's TLDR. Sorry for confusion

Strangely, it seems the etiquette would be that a TL;DR at the beginning of a post suggests a summary of the article. TL;DR at the end summarizes your comment.

Do you write this based on the article (where?), or on information from some different sources (private?)? I couldn't find such information in the article, while it mentions that e.g. "In August 2013, Mercedes-Benz [...] drove [103 km] between the German towns of Mannheim and Pforzheim without any driver input", which seems to have happened indeed (https://www.mercedes-benz.com/en/mercedes-benz/innovation/au...)

I meant my comment's TLDR. Sorry for confusion -----

Or, if you're Tesla, people pay you for the privilege of collecting data for you. [0]

[0] https://backchannel.com/how-elon-musk-and-y-combinator-plan-... "Probably Tesla will have more real world data than any other company in the world."

Interesting from a couple of angles.

First, there's this widespread handwaving around getting the very detailed map data that is at least useful for even limited autonomy. At one level, it's a solved problem that that doesn't mean it won't take a huge investment to collect all this data.

The second is the idea that this mapping data may be a source of competitive advantage for individual firms working on improved assistive driving systems followed by autonomy--initially under limited circumstances. On the one hand this makes a lot of sense. On the other hand, it means that a lot of work probably ends up getting duplicated.

> You know with current technology we can't have completely driver-less cars.

Why not? Doesn't the Google car have a near-spotless record with minimal human intervention?

I could be wrong, but last I heard google's self driving cars still required a high-precision 3D map of any route they operate on. They can drive on marked roads in mountain view and a couple other areas Google has mapped, but they're a long way from sensors that let cars drive an arbitrary route on any road.

Under a very limited set of conditions. Anyone who believes that this translates into door-to-door general driving without a human even present anytime remotely soon is... optimistic.

Living in Mountain View and driving with Google's self driving cars every day, I have to say I think they are quite a bit further than you give them credit and those conditions appear to be less limited than you are letting on here.

Something I've long wondered: howwell do they deal with inclement weather?

They do well in rain but snow remains a problem, not due to icy roads (at SXSW Chris Urmson said they do well on slick roads) but that the "better than GPS" LIDAR-based location tracking stops working when snow is on the ground.

The car simply cannot figure out where it is when there's a significant layer of snow on the ground because the reference maps stop looking like what the car is seeing.


Maybe I'm missing something but that number sounds hard to believe. An S-Class in the US is $100K. Even if a stripped down "utility" version was available at $50K that's $5B worth of cars. Yes, it's a "long term" order (i.e., not paid for and delivered at once), but that seems way beyond anything even Uber would do.

The deal is probably similar to the way an airline buys planes. E.g consider when Ryanair bought 200 737s for $22Bn [1] which is about the same in it's entire market cap.

An order like this is probably:

* For a delivery spread over 10+ years

* The vehicles will almost certainly be leased/finance

* Each vehicle will probably assume a 5+ year working life and have a residual of about perhaps 25% of list after 5 years given the likely mileage

* There will be a big initial discount given the size of the order

So assume each car, is $100k new, but given an order that size is perhaps a 40% discount. So that's a $60k sale. After 5 years, the car is perhaps worth 20k.

So the unit economics are that each car costs (before financing costs) 40k over 5 years, or $8k a year. This is a relatively low cost given the number of rides it can take and the cost of the driver who will be driving it.

On that basis is doesn't sound that expensive.

[1] http://corporate.ryanair.com/news/news/14908-ryanair-places-...

> After 5 years, the car is perhaps worth 20k.

No way. Absolutely no way. Looking online I'm seeing estimates of 300,000 to 500,000 miles per year in a taxi. If Uber is even NEAR that max, this car is going to have 300,000 miles on it. Hell even 150,000 is a ton of miles on a used vehicle.

Looking at KBB with a base S-Class that's 5 years old with 300,000 miles you're looking at roughly $12k in "good" condition (because, let's face it, people are going to be in this thing constantly and "good" is the rating the majority of used cars are in when they're sold).

Then factor in that Uber is not going to become a used car company (at least I would imagine they wouldn't) they're going to have to sell even lower than KBB to a third party so that third party can make any money. So I'd bet one of these cars would end up selling for $6k to $8k depending on mileage and condition after 5 years.

Lowering the sale price to $7k only bumps the per-year cost to Uber from $8k to $10.5k.

True which is a $250k difference if my numbers are even correct (I mean I think they're closer than parent's but who knows). But $250k isn't that much when the company is a billion dollar company.

A difference of 250m, rather than 250k, which again, I guess still isn't that much for a 60 billion dollar company. Total depreciation on these is going to eat up up a billion in value a year, or only about 2% of market cap, assuming they keep or grow their valuation.

Oh man how could I make such a mistake? I got thrown off by the lack of zeros. I can't even edit it. :(

Cars are devalued by mileage because the average car is taken care of so poorly that small damage accumulates into big, irreversable damage over time. For example, waxing wears off -> unwaxed surface pits -> rust forms -> holes form. If you re-wax the car often enough to prevent it from ever pitting, you'll never see rust or holes no matter how long you own the car. Same for most of the other components.

Individual drivers may not be able to take very good care of their cars, but large companies that use fleets of cars can take them regularly (perhaps even every night, as their default storage) to maintenance garages and have them restored.

Except that, in practice, rental fleet cars are typically only kept until they have fairly modest mileage and then sold off.

Doesn't matter how many miles are on a car, all that matters is how many are left and incremental cost per additional mile due to maintenance/repair. :)

I think this explains why they need Mercedes. It is a nice car that can have a lot of mileage easily.

The average taxi puts 70k miles per year on their car. So after 5 years these will have 350000 miles. Its not going to be worth anywhere near $20k

Great analysis but this makes it sound even less likely. Going from zero marginal costs to something like this seems totally against their current model.

It seems like it would make more sense for a partner company to be the one that has all of this in the books and Uber just leases from them.

so they pay 800,000,000 per year for 5 years?

$40,000/car * 100,000 cars = $4B. that's only a 20% discount over the course of 5 years.

In the states the franchise laws do not allow Uber to buy directly from the manufacturer. Also, the margin on an S class sold in the US is approximately 15%, not 60%.

Thanks for the breakdown, it was a helpful analysis.

"Uber had placed a long-term order" [...] "Uber is particularly interested in autonomous driving vehicles, the magazine reported, adding that such cars are expected to be available after 2020."

Perhaps the long-term order is a commitment from Uber to motivate Mercedes to invest more heavily in autonomous vehicles? If the long-term strategy of Uber is to have a fully autonomous fleet then a long-term commitment of $5bn with one manufacturer seems plausible.

I'm sure there's a volume discount when you order 100,000 of anything, even an S-Class. There's also the possibility of this not entirely being a cash sale.

That is, if the report is even true.

Yeah but how much of a discount? The example in question was already assuming $50k per car which is already a tremendous discount. Even if you cut it in half again to $25k (which there's no way the price per car is that low) that's still $2.5B in cars. That's insanity.

Depends on how this is structured, if it's say 33k cars on a five year lease, renewable twice that's ~2.5B over 15 years. Also, Uber has a huge and arguably inflated market cap, so if they get this deal in exchange for stock it could be very valuable to them in the short and long term.

Which is why I suspect this isn't a simple cash transaction, and may involve equity or revenue share or something along those lines.

Think so too. Most likely Daimler will get a stake in Uber - which makes sense for both sides.

Uber is what you need to make an auto-driving automobile acceptable to the general public. With Uber spin, Daimler can push on the driver-less car in many markets..

Sci-fi days ahead methinks.

Tesla seems to be doing a fine job showing the benefits and safety of autonomous vehicles to the general public and regulators.

Which Tesla vehicle is self driving?

Both the Tesla Model S and X are Autonomous Level 3 capable:

"The driver can fully cede control of all safety-critical functions in certain conditions. The car senses when conditions require the driver to retake control and provides a "sufficiently comfortable transition time" for the driver to do so."


Its expected their Model 3 mass market vehicle will contain the same, if not a greater level, of driving automation.

Disclaimer: I've traveled down the interstate in a friend's Model S for ~2-3 hours at a time at ~85mph in autopilot mode with no intervention required.

Why would it be in Daimler's interest to push driver-less cars? Their sales would go down in the long run.

Long term, driverless cars are going to happen so they might as well get in early.

If you can't beat them, join them.

Especially if there is no middle-man dealer to pay. On a $100k S-Class, Invoice comes in around $93k. Some people are able to pull off sub ~88k sale prices at dealers.

I wouldn't be surprised if the wholesale (with profit) cost on a S-Class Benz is somewhere around 65-70k. I'd imagine that the cost of building an S-Class (100k retail) over a C-Class (~40k) is nominal.

Respectfully, your inference is entirely off base. Trivial research gives you the wholesale value. A slightly deeper dive gives you the manufacturer margin. Franchise laws prohibit direct sales.

Well, if they expect to use them as self driving cars (which the article mentions), the investment can pay out pretty quickly. Say they can make on average 5$ per ride. To make $60k they need 12k rides. A self driving car should be able to easily make 12k rides in one year (only 32 rides per day).

Only 32 rides per day? I would think that's pushing the limit. You will have to buy cars for peak times, so most of them will see only a few hours of use each day. At a guessed 6 hours of busy service a day, that's 5 trips an hour, or 6 minutes to drive to a customer and 6 to drive him where he wants to be.

Also, a risk here is that other large parties will make the same calculation, flood the market (anybody with money will be able to start a cab company with a lot less hassle from labor laws because you don't need that much personnel) and drive margins down.

To win in this space you probably will need both a lean model of operations and a good image in the customer's eye. Uber currently seems to have both, but committing _now_ to a provider of cars is risky (what if other manufacturers turn out to have better or cheaper self-driving cars?). On the other hand, it also may be necessary to commit now in order to stay lean (committing now probably gets them a nice price cut)

Perhaps they're purchasing these to handle the constant/baseline load of riders throughout the day, and then will continue to use human drivers & contractors to supplement demand beyond that, eg. the peaks times you mention.

So it'd be the best of both worlds for Uber: being able to put the self-driving vehicles to use around the clock and not having to make a capital investment to handle spikes in demand, which would just continue to be handled by existing drivers (for now). Just call in the human backup to help out when required.

I'm also wondering if they made the huge order so that they'd be able to influence and guide the requirements for the vehicles to Uber's specs. It seems that the requirements for a self-driving car for an individual's use differs from that for commercial taxiing.

6 minutes to drive to a customer and 6 to drive him where he wants to be.

You're assuming that each ride is independent, but that's already not the case with their Smart Routes - the driver picks up a customer, then picks up another one on the way, then drops off the first and picks up a third, etc.

The goal stated by their CEO is "perpetual rides", in which the car is never without a passenger.

Perpetual rides may be their goal, but I don't see how they will not get there. For example, there's no way they are going to find as many people wanting to go to sport stadiums as are wanting to get away from them, just after the end of the game, and stadiums typically are so far out of about every other destination that they cannot make a detour there with a paying passenger on board to pick up a second passenger.

That applies to a lesser extent to such things as theaters and cinemas, and even suburbs (who wants to get picked up there at 1 AM to fill the car that that theater goer took home?)

They will get below 50:50 for "at least one passenger on board", but it would surprise me if they managed to beat a 25:75 ratio.

>> "...what if other manufacturers turn out to have better or cheaper self-driving cars?"

Cheaper, certainly. Everyone will have cheaper SD cars than Benz, just as everyone has cheaper cabs than an S-class today. But better? If history is any guide, German luxury will be the gold standard of automotive UX for our lifetimes and beyond.

I don't think they need to tackle peak driving time at the start. They still have all of those regular drivers. Surge pricing should still draw out humans.

That was my thought actually - but reversed: Why on earth would they need that amount of cars when self-driving cars are just around the corner?

If these are self-driving cars, it makes sense to me, otherwise it doesnt.

It's hard to come up with a huge number of cars on the spot once self-driving cars are available. It's much easier to retrofit your already existing cars.

Retro-fitting. They buy plain old cars and then slap their sensor suite on top.

That seems way harder unless they plan on buying a very small number of car models. Every different car model (even the same model between years) has different performance characteristics (do you issue the same orders for "hit the brakes" to a car that needs 100ft to go from 40-0 as a car that needs 50ft?) and physical dimensions which would necessitate different software optimizations and sensor calibrations. Development plus rigorous testing of every configuration means this quickly gets out of hand unless you standardize on 1 or a very small number of models.

Hence buying 100k of the same car.

The deal could be structured like contracts for planes.

Uber placed an order for 10,000 vehicles with the option to purchase 90,000 more.

That's what, 1/9th of Uber's current valuation? And it's the heart of their business.

"it's the heart of their business" but the fact that it's a fundamentally alternative business model from anything they've taken a cent of revenue on is important to note.

It's like if airbnb bought $5B of Manhattan real estate. Yes, it's "the heart of their business" but also - wow.

Sure - but I think it's always been the expectation that Uber/Lyft/Sidecar before it went bellyup were participating in the sharing economy only as a stopgap until autonomous cars are available.

With Airbnb, I don't think anyone expects them to pivot into owning their own property. There's no real benefit.

I don't know, if I were an investor in Uber and saw them place a $5B order on Mercedes S-Class cars, I would think that there's a good chance they're worth $100B.

Haha, I just checked. A recent valuation is $68B. (Dec '15)

Agree. Even if Uber got them at 1/2 price, let's say $60,000, the math doesn't add up. Assuming they don't do some fancy accounting and tax magic.

Or something completely boring like leasing them.

Over ten years, that's $6k a year.

Yeah, I think you are missing _something_ ;)

I am reminded of Lucent, who built the best hardware for the end-user ISP market in the form of the Ascend MAX TNT, which would terminate a DS3, break it into individual dialup lines, and handle all of them as modem calls. Cool stuff.

It's just that they sold all this stuff to ISPs who didn't have any good credit lines, so Lucent extended them credit (bing! interest payments!) using the MAXen as collateral.

When the ISP dialup market collapsed, the ISPs returned the equipment in lieu of payment. That equipment had no value whatsoever, because anybody who wanted one and could afford one already had two.

Presumably there's some sort of secondary market for cars -- but a hundred thousand black S-Class is about what they've sold in the US over the last 7 years combined.

Risk = Reward. If UBER doesn't invest in transportation powered by automated cars they risk being replaced by companies that do. General Motors and Lyft are already ready to do it if UBER doesn't. http://www.wired.com/2016/01/gm-and-lyft-are-building-a-netw...

Unlike the days of dial up where bandwidth improvements made it obsolete it's hard to see a day where we don't need reliable transportation on roads. Cars play an integral part of modern life.

The maintenance cost of these cars in the US is high. Whoever is making this deal should probably stop by the Lexus dealership before they sign any papers.

> The maintenance cost of these cars in the US is high.

You are correct, of course, but I think the majority of that would go away if you hired the mechanic yourself and bought parts direct. My theory is that most of the German luxury vehicle surcharge is the "hey, it looks like you've got money, how about you give me some of that" surcharge.

I mean, I've owned a BMW 325is and a Toyota Sienna. Similar vehicles in that the vehicles had similarly sized engines/transmissions and brakes to match.

Sure, my experience agrees with your assertion in that rolling into the mechanic in the 325is is going to cost me north of twice what rolling into the mechanic in my Toyota Sienna will cost me.

But if you are doing the work yourself? the German car isn't any harder to work on. If anything, I think the german mechanical aesthetics agree with me in ways that make working on it easier than working on a Japanese car. Sure, they need a little bit more love in that they're just not as advanced as a Japanese car, or at least not as much effort is put into maintenance free operation, so you need to do more to keep them running, but that difference isn't huge. Parts are also not hugely more expensive than similarly specc'd parts on the toyota (e.g. brake rotors cost about the same on my 325is as they did on my toyota sienna; they were similarly sized, too; it's just the 325is was a much smaller car, so stopped dramatically better.)

I mean, I'm generalizing about German luxury vehicles from my experience with another brand, but yeah, if you hire your own mechanic, I bet most of the maintenance cost differences become much smaller.

I once asked my father (a lifelong auto mechanic) why servicing high end vehicles is so expensive. Roughly from memory, don't quote me on makes/models/etc:

Most cars require special, often model specific, tools to service them. A specific adapter to take off an oil filter on your Honda Civic, or a set of special ramps that get your super low Ferrari on to the car hoist. The shop then has to amortize the cost of tools over the number of vehicles they service with it. So the special wrench used for the Civic can probably be used for 500 oil changes in a year, but the Ferrari ramps are used on maybe 10 a year. Not to mention the specific training required to do the actual service.

In a computer related example, imagine a receptionist that just needs Microsoft Paint to put text on signs around the office. Now compare that to an industrial designer that needs a $10,000 licence for a CAD application and specific training to do material stress analysis on your new iPhone accessory. Each sign has little to no incremental cost, whereas each iPhone accessory needs to pay for part of that licence and class.

If Uber were to buy thousands of the exact same vehicle, they would have a predictable amortization of a spark plug tension wrench, while a dealer might hope to get enough cars in a year to justify the cost.

I'm a mechanic and software guy. Your theory is incorrect. German parts and labor generally cost more. Mostly due to complexity. An S-Class has more on-board electronics than the average Boeing. They begin to have issues after the 2 year mark and are not cheap to fix. Even if you have your own shop and parts. The 325 you had is a simpler BWM that did not carry such complexity.

I'm not saying you are wrong; the new cars might well be way more complex in ways that might cause people to spend more time troubleshooting, I wouldn't know.

I haven't dealt with anything really more complex than the early 'aughts toyota... at least up until that point, the electronics made it way easier for a shadetree mechanic like me to solve problems. There's all sorts of ways to get diagnostic codes out of the thing; on the bmw, you did this dance where you turned it on, then floored the gas a certain number of times, and then the 'check engine' light would tap out a code you could look up in the manual, just like beep codes on a motherboard. On my Toyota, I hook up the OBD thing and it gives me a code I can plug into google and get youtube videos of how to fix it. I've never dealt with an electrical problem more complex than one of the the (what do you call them? the short wire things that go from the coils to the spark plugs on the bmw) going out of spec in terms of resistance, but that was pretty easy to diagnose with the manual and multimeter.

But if it really is the complexity that creates the cost, why was it so expensive to pay someone else to service my simple 325is?

Factors such as location, brand, and demand are important. Fixing BMWs is generally more complex than fixing a comparable Japanese or American model due to specialty tools, software, and complexity. Their wiring system is closer to what you'd find on a data center than what you'd find on the typical car.

I currently still turn wrenches as a hobby and work on higher end Euro cars (Ferrari, Porsche, etc.) and it only gets worse. :)

Lexus is nowhere near Mercedes

Absolutely. One is headquartered in Germany and the other in Japan/USA. Lexus cars are more reliable, though.

There's a market for used taxicabs and police vehicle. It's not a very vibrant market, but a car with a working engine will fetch something before being considered for its scrap value.

The S-Class is an interesting choice. I'd wager that most people have never been in a car as nice as an S Class before -- no comparison really to the town car people associate today with Uber Black. The back seat has massage, heat and cool, individual entertainment options, power recline. Some of these things could be customized, but Mercedes isn't going to tarnish the reputation of their flagship sedan so Uber can have a more utilitarian taxi.

A great reason to charge more money!

The S-Class is a very common model for a taxi in Germany.

Yes Mercedes taxis are common throughout Scandinavia also but they are workhorse models, still very nice but without the "S" level of luxury details and engine power.

In my experience, the E class is far, far more common.

No it's not. Most of them are E-classes.

Quick fact-check, both E-Class and S-Class are common in Germany: https://en.wikipedia.org/wiki/Taxicabs_by_country

I am from germany. I have never seen a S-Class taxi.

I've actually seen a Porsche Panamera taxi in Germany.

Something doesn't have to be the majority of things to be considered common.

How do taxi drivers afford them?

Unfortunately I don't have any articles to back this up, just anecdotal discussions with drivers in Germany, but;

First: they're a domestic car manufacturer, lower cost to deliver to local drivers. Second: they're able to offer specific taxi options, such as manual front seat adjustment (vs automagic). Third: they offer a substantial discount to taxi drivers (they justify the discount through the marketing value of having large number of individuals ride in their cars).

Such discounts for marketing purposes is quite common in the industry. I've also heard dealerships sell cars to rental car companies at a loss, just to make sure customers have access to their cars (and will possibly consider buying that type of car for the next purchase).

Would you have any idea what the cost to build an E class might be? Just wondering how the various taxes around the world affect its affordability.

In Denmark, you could buy a car for taxi use at a much lower tax rate than normal people, but after driving it with passengers for around 250,000km and 2 years you could sell it as a private vehicle. The calculation made the car extremely cheap (I've seen a calculation of deprecation of 20% over 2 years despite the heavy use ) -- so you might as well get a nice Mercedes.

The tax a private person has to pay to buy a car was 180% in 2015 so essentially a luxury car is nearly triple of what it might be in US, with a new S350 being $250k.

Thos taxi rules have changed recently however, so Mercedes may lose their 90% taxi market share.

I've always curious about why the taxi differences between North America and the rest of the world. When I was younger in Asia, we had these "luxury" taxis that were still very reasonably price. I was appalled that taxi rates in North America were so high and the cars were just terrible.

Thanks for your clarification!

Another concern for uber is what are people going to do to the inside of these driverless cars now? Passengers today are respectful to ridesharing vehicles they sit in--because the owner of the vehicle is close at hand and the human connection commands some respect. But when its just a group of (maybe drunk) passengers in an empty vehicle--where's the accountability?

Passengers won't be anonymous and Uber already has reputation system. Also you can already rent a car, and rented cars are usually not vandalized.

I imagine it'll be kinda like car2go or ZipCar where if you get in a car and it's messy or damaged you're supposed to report it. Presumably the person before you did the damage. If you don't report it you risk the next person reporting it and you get the blame.

Car rentals are fairly expensive. People tend to take better care of things they've put a lot into. Case in point, rental cars tend to be in much better condition than the back seats of taxis, which in turn are in better shape than the back seats of buses. I think people's concern is that single rides in self-driving cars might even rank below buses.

Still, Uber knows who you are. If you damage their car they can invoice you. Presumably they will have CCTV in these things?

Rental cars aren't vandalized because they are inspected before and after every rental by rental company employees.

Driverless rental vehicles will very quickly look like the back of public buses. "Dave {hearts} Christina" inscribed on every surface.

Have you heard of Zipcar? There are lots of reasons to be skeptical that we'll see autonomous taxis in less than decades but this doesn't seem like one of the bigger problems.

A car sharing company accused me of leaving a dirty vehicle. Apparently I had picked up smoking and gotten a dog without knowing it. They were quite adamant about charging me a cleaning fee until I got American Express involved.

Presumably there were one or more intervening drivers between whoever caused the issues and you who didn't notice or didn't think enough about it to complain. (Both the things you describe are probably way more irksome to some people than others.)

They are driven carelessly, smoked in, eaten in, etc...

You have a lot more liability as a renter.

Cameras? You could even just take a picture when they come out and when the cars come back for the cleaning crews, they'll find out pretty quickly which rider to charge the cleaning fee.

Same thing we do for car co-ops I guess (next people report the car is fucked up) -- plus cameras?

Uber has some of the best researchers in the world. I'd bet Mercedes builds the cars, specialized with Uber specified additional sensors and hardware for self driving. Uber customizes the source to deal with tough driving conditions, and turns the code back over to Mercedes.

Uber gets it's fleet, Mercedes gets the tech. It's getting competitive. Mercedes will get the tech perfected, but this kind of deal could get them there much faster. Access to Uber's team, and more importantly mountains of data that Uber is taking the risk for will probably speed up refinement of the tech.

This has to be the answer if the rumors are actually true (the article does say one source said this, one source said this wasn't happening).

All auto companies need to get into the self driving space. Uber wants to get into this space for transporting people. If Uber and a car company work together Uber could get a discount, help develop the tech and share with the car company; it's a huge win for both sides.

At least that's how I'd try structuring the deal. You know, from my billion-dollar-company-CEO-arm-chair.

Uber is an interesting company for a number of reasons, but here's the part that interests me the most:

Uber operates in a grey-area legal environment in the cities it operates in. Uber's defense / view on this is that they generate such a societal positive in terms of jobs that it outweighs the legal constraints.

But, when your goal is to eventually use a fleet of self-driving cars and eliminate the job possibilities for drivers, isn't that a moot point?

> "über's defense/view on this is that they generate such a societal positive in terms of jobs that it outweighed the legal constrains"

I doubt they care a bit about the amount of jobs they are creating, I would say their view is "they generate such a positive amount of cash that it outweighs the legal constrains" and if they see a possibility of removing their highest expense of getting that cash - they won't bat an eye on doing so because they already have a history of doing stuff that is on the edge of legality/morality.

I think it will be very interesting to see what happens with self-driving car regulation. I foresee the money going from taxi drivers -> Uber -> Gov/Uber Split & no more jobs.

I don't think those are Uber's main defenses, which I think are actually:

1) They fall under the traditional pre-arranged/limo exemption since they aren't doing street hails.

2) They massively increase the convenience and affordability of getting a ride, which cuts down on drunk/bad driving and saves money.

Neither of those points is affected by the move to SDCs.

When that happens, skirting the law will no longer be necessary. A company with no personnel costs, rather than low personnel costs, should have little trouble cutting a more attractive deal with municipalities. Or they can just lease the technology to the existing medallion-holders.

> But, when your goal is to eventually use a fleet of self-driving cars and eliminate the job possibilities for drivers, isn't that a moot point?

The fetishism of "job-creation" is misguided and unsustainable. The purpose of life is not 'work'.

High-availability self-driving Ubers are undoubtedly better for the majority of urban consumers than taxis or individual car ownership.

They aren't creating jobs thou? They are creating a cheaper/better car service?

This article is a pretty good discussion on your exact point:


"Back in September of last year, Kalanick said at a conference that Uber was creating 50,000 new jobs for drivers around the world every month—more than doubling the 20,000 new monthly jobs Uber estimated it was generating in May.


Usually when we talk about job creation, we’re thinking in terms of the Bureau of Labor Statistics’ monthly payrolls report. So the economy added 126,000 jobs in March and 264,000 in February, while the unemployment rate was basically unchanged. That’s the language of the labor market. Because the drivers who use Uber’s app to accept your request for a ride and pick you up are independent contractors, they fall under the BLS’s umbrella of “alternative employment arrangements.” "

Mercedes/Audi/BMW taxis are common in Germany. This doesn't mean you'll be able to order your S class in NYC/SF any time soon.

You can get an S Class or similar via Uber today, but you have to order an Uber Lux, which is quite a bit more expensive than a regular Black (though I have gotten a BMW 7 Series as a Black before). Of course, none of these cars are owned by Uber.

My thoughts exactly. Uber are struggling to get a foothold in Europe because of regulation. I suspect the plan is to become a traditional taxi or limousine operator in these hostile markets, leasing cars to licensed taxi drivers. Using S class rather than the more common C class cars would fit with their existing Uber Black service.

> in these hostile markets

"markets where regulations are enforced"

Have they piloted this idea? Because 100,000 cars is big bet if they haven't established service-market fit.

Taxis are not S class in Germany.

Here in Berlin, they're E class. So you're talking about E-segment vs F-segment cars here. Not that big of a step here.

And most of northern Europe

Here in Norway we even have Tesla Model S taxis. (The government subsidies are a bit crazy.)

Edit: to be clear, these are regular taxis, not high-end/limousine/whatnot-service

I've seen it claimed that due to the lower fuel costs, and the lower maintenance, that taxis are a fast growth market for electric cars so that may be a factor as well.

Are there actual subsidies or just lower taxes on electric cars in Norway?

EDIT We are being pedantic apparently. Are there any cash or cash like payments or is the subsidy purely in the form of a tax reduction?

Most of the subsidy is in the form of tax reductions, a little is in cash like payments (e.g. increased car allowance). But what does it matter? The distinction is purely arbitrary. If the government said "there will be no tax reductions, you get an equivalent cash payment instead", the end financial situation for both you and for the government is exactly the same.

If I sell you a donut for $3 and then give you $1 back, or I sell you a donut for $2, is there any meaningful difference?

Investopedia defines subsidy as

"A subsidy is a benefit given by the government to groups or individuals usually in the form of a cash payment or tax reduction."

Southern too. There are tons of Merc taxis in Greece. Beemers or Audis not so much.

This is the (German) article this is based on:


It says they plan to buy "a six digit number of cars", and they're only interested in self driving cars.

It Daimler can deliver cars that are able to handle Uber rides without a driver, including driving empty to location (considering Uber would send a destination location) then 100k vehicle does not sound like a lot. That feels like a big if, though.

Having a deal in stock rather than cash also makes sense: it is unlikely that many people would buy cars from Daimler or anyone if Uber charges at cost (gas + maintenance).

no pressure on Daimler then...

It's like a Kickstarter, but with several billion dollars.

How does this benefit Uber?

Imagine Daimler can produce 100,000 self-driving cars. They would continue to produce and sell self-driving cars to other people as well. They could sell self-driving cars to Lyft. Individuals could operate their own self-driving car services. They could organize in to cartels and have a single app that competes with Uber or Lyft.

Uber is destined to be a commodity in an open marketplace of taxi services as soon as they adopt self-driving cars. Their primary roll right now is as a labor organization. They currently create the economic incentives needed to attract a fleet of drivers. Self-driving cars are not motivated by economic incentives. Uber can't control the sale, distribution and organization of self-driving cars.

All that riders care about is being able to reliably get from one side of town to another. They'll know that they can trust Daimler self-driving cars, regardless of the dispatcher.

We're going to end up right back to where we started, with a commodity system of private taxi dispatchers and a productive industry of automobile manufacturers, with an emphasis on the automatic nature of this new form of transportation.

Daimler is the big winner here if they can bring a self-driving car to market.

Uber's (differential) advantage is not so much in labor organization, but rather, in having already built out a customer network and having solved all the logistical and (passenger-side) UX problems that are specific to this kind of application -- how to store, retrieve, and exploit the data, what UX decisions make the app easier to use, how to manage support, etc.

Other entrants (Tesla or Mercedes) will have to reinvent this wheel and have to market themselves and get users to download/use yet another app that does the same thing.

Personally, I think the right move for Tesla/Mercedes is to just rent out SDCs and focus on their core business rather than expand out into a different one they're new to.

(That's not an advantage against Lyft, of course.)

I think the other function Uber serves is not just an aggregator of supply (of cars and drivers) but of demand (passengers). It is far from a trivial effort to have a sufficient amount of users know of your app/service, download and install it, trust it enough to add payment details and use it.

While not quite as insurmountable an obstacle as, say, making a new smartphone platform (sorry Palm/Nokia/BlackBerry/Microsoft), it is enough of one to make this far from a perfect, liquid market.

> I think the other function Uber serves is not just an aggregator of supply (of cars and drivers) but of demand (passengers).

I think you misunderstand Uber. Most of its perceived value is in arbitrage between drivers' perceived and actual costs of car ownership: drivers mostly think in terms of wages minus cost of gas, ignoring all of the other costs. Given that business model, I doubt Uber is actually looking at buying these cars. They will presumably force drivers to buy or lease the cars themselves.

So do individuals buy or lease self-driving cars in the future? Would they then set up their self-driving car to take instructions from Uber's central dispatch service in exchange for a cut of the profits?

Currently both Uber and Lyft handle the task of vetting drivers and making sure there is a good experience for their customers. They organize and control the labor required for a good experience for the riders.

Think about a rider in a world where Lyft and Uber are both dispatching self-driving cars. What's the difference between the services if they're all using Mercedes S-Class SelfDrivers? The rider knows that all Mercedes S-Class SelfDrivers are the same experience. There's no question of if the driver is trustworthy or competent. Uber and Lyft would be providing the exact same commodity experience of sending a self-driving car to your location to then do your bidding in exchange for money.

> So do individuals buy or lease self-driving cars in the future? Would they then set up their self-driving car to take instructions from Uber's central dispatch service in exchange for a cut of the profits?

In Uber's perfect world, yes. And Uber will do its best to hide the true costs of insurance, depreciation, regular maintenance, etc. from owners and lessors, just like they do today. It's worth understanding that Uber only works as long as it doesn't own any cars, or employ any drivers. If they switch over to owning some self-driving cars, they'll have to maintain and insure them.

Customer acquisition is, I mean, it's not like you literally just snap your fingers, but it's very straightforward. There are hundreds of channels to acquire customers, and services to help you do it.

In a hypothetical world in which we have fully autonomous vehicles and someone wants to compete with Uber, anyone who can afford capital investment in a fleet is not going to have much difficulty generating a customer base (and then they can have a mutually destructive price war with Uber).

It's possible that Uber & Daimler are no longer completely separate companies, right? One of them could now have a significant stake in the other.

The magnitude here is pretty impressive.

I found this list of sizes for existing commercial fleets. They would be besting UPS for total fleet size if all 100k were on the road at once. http://www.fleet-central.com/content/pdf/AUTOF_top300commerc...

This makes a lot of sense. S-class cars already have the sensors for self-driving in highway conditions. Uber UATC is actively working on the software for autonomous drivers.

This is a large scale capital investment in a safe luxury service that Uber can sell in certain markets. I'm betting the s-class needs very little hardware to wire into what UATC is developing.

Anyone know how this volume compares to a Mercedes dealer's volume over 5 or 10 years?

Large dealers in california sell around 1200 S classes per year. More common dealers sell between 200 and 300 per year.

Or were you comparing that to total volume? Market leaders sell a bit over a thousand per month.

Awesome, thanks, that helps bring some perspective.

Why would Uber want to buy non-self-driving cars? Right now, they have their drivers buying the cars for them. Their current business model shoves all the capital costs down onto the driver. If Uber owns a lot of cars, their business model looks far different, more like Hertz. This will reduce their valuation substantially.

Yes I wonder if this is a little short sighted. Currently Uber needs to hire drivers who use their own cars. Once self drive vehicles become available they only need the cars. Given that most peoples cars spend the majority of their time parked, waiting to be used, that is an awful lot of potential cars for Uber to rent. I'd be very interested in having my car earning money while I'm at work or asleep.

Presumably they believe the savings from being able to stop paying drivers will more than make up for it.

Because they are going to lease them out to drivers to compete with Lyft, which earlier this week announced a similar program.

Uber already has a financing deal for drivers.[1] This may just be a mod to that plan.

[1] https://get.uber.com/cl/financing/

I think it's a response to Lyft, whose plan is more convenient because it allows drivers to rent the vehicle.

Probably looking into the future which will likely be all taxing services will be self driving.

For the sake of comparison: the three largest US rental companies: Enterprise, Hertz and Avis together own about 2 million cars, have about 15,000 locations (most of which are competitive with each other), and have about $25 billion in annual revenue.

Why do I bring up rental cars, a market usually considered distinct from taxi and limo services? Because without a driver, there's no difference.

Uber is trying to crush the taxi market, but in an age of self-driving cars it needs to watch out for the rental car giants who will happily extend their daily-weekly rental cycles down to per-trip rentals, with pickup and drop-off anywhere in the country. The giants already have purchasing arrangements with the major auto manufacturers, supply chains, maintenance facilities, and parking lots to use as buffers.

Someone (their insurance? Them?) Won't let you rent until you're 25.

Very telling that Uber is trying to go with Mercedes, when Tesla is really their first choice. It's an open secret that Elon Musk will be launching an Uber competitor using Tesla vehicles. I expect a "surprise" announcement from Tesla within 18 months.

The other thing is Tesla can't make this volume of cars yet & it carries a certain amount of risk for such a large deal.

They only made about 50,000 cars last year and have capacity for around 90000 this year. Even if Uber's order was spread on a number of years, Tesla will have trouble filling the order.

At that scale, customization is a given. Uber could leverage their own robotics resources together with Mercedes' to speed up autonomous vehicles time to market which Uber stands to benefit from substantially enough to justify this kind of investment.

At that kind of volume, would it make more sense to have MB develop a bespoke car for Uber?

The S-Class is a great car, but I'm sure there are Uber-specific improvements that can be made. Why would you order some many of something 'off-the-shelf'?

I don't think you'd find any manufacturer who would tool up for a bespoke car. Uber either needs to buy an existing manufacturer with existing large scale capacity, or build out their own capacity. And we've seen with Tesla how much and long that capacity takes to build out for only 100K vehicles/year.

Uber is capital chasing a potential treasure chest. Google, Tesla, GM/Lyft/Cruise are the real players it appears.

There are contract manufacturers also in the automobile industry. For example the Finnish Valmet Automotive has been building Porches, and is now doing some MB cars.


A quick Google search didn't find me their manufacturing capacity, but I'm curious if it approaches what Uber might need to compete.

They might do a special trim package for that volume.

That's one way a precommit makes sense to me; a contract could be structured such that Uber gets a certain price as long as they keep up with their commitments, if they don't keep up, they pay a penalty.

I could see a trim package, but if autopilot hardware is only a few thousand dollars, it would behoove Mercedes to put it in all of its vehicles. At that point, any startup could build an interface to allow any Mercedes owner with the hardware to lease out their car through a P2P ridesharing platform.

Bye Uber.

People said the same thing about Microsoft.

As it turns out, going from sensors to working self-driving car is a pretty difficult task, and Uber has some of the best self-driving researchers in the world working for them.

Right, but Tesla has vehicles on the road, now, with a high level of autonomy (driving across the country autonomously 96 percent of the time: http://jalopnik.com/we-set-a-cross-country-record-in-a-telsa...) that are gathering massive amounts of data (40 million/miles/month) from vehicle owners as they drive.

Its an uphill climb for Uber. They will need to fund their vehicles; Tesla funds their battery manufacturing, vehicle manufacturing, and autopilot telemetry corpus buildouts using the vehicles owners have paid for.

1. I was responding to your Mercedes/Uber scenario; the comment I was replying to literally didn't mention Tesla (which fits more aptly to Apple's position in the desktop computing analogy).

2. Interstate driving autonomy and city driving autonomy are different beasts. Gathering a ton of data about task X isn't necessarily helpful for task Y; ML isn't magical. If anyone, Google is ahead of the pack in the market Uber is vying for (automated taxis).

100k vehicles/year is 1/2 to 1/4 of capacity of a vehicle production line; any automaker would retool for 100k guaranteed sales, if the price was right.

If this is true, this is an incredibly risky move. I know self-driving cars are more or less inevitable at this point, but there are hundreds of unknowns that still need to be figured out. How will passengers treat the vehicles? What happens when there is a problem, and there isn't a human present? How secure will the cars be? Is the 'call a car from an app' business model the one that will win? Committing this amount of capital this early in the game seems like a big gamble, unless there are a large number of contingencies worked into this deal.

>How will passengers treat the vehicles?

As a mobile drug/prostitution marketplace.

Wonder what level of surveillance these will have. Surely that would be an excellent data mining opportunity.

It's probably more along the lines of: "Whenever you guys are done making it fully autonomous, we want 100,000 of 'em to replace our drivers"

If that was the case, the volume makes sense. Makes zero sense to buy 100,000 of today's model and watch it depreciate in value while providing none back 'cause you still need a driver.

I'm kinda curious how it's possible for an organization to go through all it takes to produce self-driving cars (engineering, qa, logistics, sourcing, etc.) but is unable to build an app to find riders for these cars and manage yield of available seat-space.

Not really seeing the value-add of Uber at that point beyond the brand and historical data about customer rides as transportation is so fungible.

It's not entirely impossible that Uber/Lyft will engage in a race to the bottom while a bold automaker (such as Tesla) offers to buy out Lyft for it's trip data... to then stand on the sidelines as Uber goes bankrupt... to then purchase their data during liquidation.

I could be wrong, but it will be interesting to see where Uber executives take the company.

I am not surprised if Uber is going in this direction. In fact, looking at what Uber has done so far , Uber has already planned for this sort of future in 10-20 years.

Look at what Uber has done:

* Uber buys Nokia Here Maps for their map part and for their street imagery part.

* Uber hires robotics department.

* Uber knows today's business is supposed to be the transition phase to going into a zipcar like car-on-demand service.

There aren't too many data points here, but the only thing I can glean from this is Uber wants to have robotic cars to pick you up and drop you off and will plan that future by storing street imagery, instead of that LiDAR based method the others are doing.

In the German source[1] it is very clear that the order is under the condition that Mercedes delivers an autonomous self-driving vehicle.

It is also clearly stated there that Daimler is not expecting to be able to deliver such a car before 2020.

Basically, Uber said: "Once you're able to build a self-driving S class, we will buy 100.000 of them". And Daimler said: "We will come back to it."


Wow, I wonder what a contract like that looks like, and how many years the payments and deliveries are spread out over. Assuming a massively-discounted price of $25,000 per car, that comes to $2.5 billion!

The lowest I could imagine them selling for is 45k per car. 25k isn't a discount, it's a loss.

If Uber really is buying 100,000 cars, or anywhere near that number, it totally undermines their argument that they are a ride sharing app and not a straight up taxi/limo company.

At that quantity, an S class is the badge on the trunk, otherwise whatever Uber wants it to be. Likely the technology for automation, but maybe not all the luxury bells and whistles.

The critical question facing our driverless future is liability. Who owns the risk? If the individual owner is liable, you can imagine the upshot - slow adoption rates, high premiums. If the seller/programmer is liable, expect consolidation. It feels like this is the question of our time. To pay for consolidated services - spotify, netflix, uber, Bernie Sanders - or to remain an individual and accept inefficiency as a natural cost of freedom.

Just a shower thought, but if the car makers can make self-driving cars, can't they also implement their own versions of Uber ?

Why would they even sell the cars, when they can rent them out and make much bigger margins ?

Uber's current strength is it's database of drivers, but since we're talking self-driving cars here, that become irrelevant.

It's not like Mercedes or BMW lack brand recognition in the transportation space...

It is actually suggested in the article: "Earlier this week Mercedes rival BMW said it was considering launching its own ride hailing service in what would amount to a rival business to Uber."

They should make sure that in the contract it says the cars are virtually unhackable, too, or Daimler-Benz is responsible for any liability from security breaches:


I hope they use these for more S-Class UberBlack service. I wanted to get driven in one for our anniversary but found it difficult.

The S Class has been Mercedes platform for autonomous driving cars. The current generation actually has a very capable installed sensors for autonomous mode. Per https://m.youtube.com/watch?v=fhDqgh6DClo

"Loss-making Uber would make drastic savings on its biggest cost -- drivers -- if it were able to incorporate self-driving cars into its fleet."

Umm, don't they get the cars for free now? How many rides will the cars need drive to break even compared to cheap drivers and no capital cost now?

They still have all of those regular drivers,It's hard to come up with a huge number of cars on the spot once self-driving cars are available.

That's huge order 100K is what they sell in a year. So if spread over 10 years this would mean they are buying 10% of S-class cars produced each year.

Wow, that's a huge amount of capital to spend upfront. Wonder if this is for their black service or more targeted in Europe.

What gives you the impression it would be an upfront expenditure? My guess is it would be a 5-20 year deal and likely include financing.

The article specifically mentions this is a long term order. So it's likely on the scale of years, and not upfront

I doubt they're spending a single cent up-front.

I admit I didn't RTA, but people keep saying things like, "Uber had placed a long-term order"...

If/when someone develops fully reliable, fully autonomous cars, what makes them need Uber?

They wouldn't but that's like saying computer manufacturers don't need the OS manufacturers anymore. They wouldn't necessarily have to have Uber but Uber providing the payment, route planning and logistics and someone else providing the vehicle seems like a pretty good match.

Though I'll admit the self driving vehicle delivery is probably another level of difficulty than Uber's current capability. But I suspect Uber and the car manufacturer may work on the self driving together.

Uber is positioned to be the platform for transportation as a service. In their ideal world no one would own a car. Owning a car, like a plane or boat, would be come a luxury item.

Possibly so they don't have to buy one themselves? Similar to Car2Go.

One of the weaker lead quotes I've ever seen.

Wait, is the S-Class even a legit SDC?

self-driving Uber cars? What could possibly go wrong?


There comes a point at which the helpfulness of technology becomes a form of oppression: walled gardens, predictive services making the wrong predictions, and every social platform forcing us to use our real names. It's cute when it's small, but what happens when self-driving cars can collude with the cops?

https://motherboard.vice.com/read/one-star ( previously: https://news.ycombinator.com/item?id=11149653 )

Downvotes? Lol. Start reading some oldschool science fiction: those are to point out possible outcomes, both bad and good, of technical evolution.

Self driving cars in our current world, in our current system will not go well. We need changes in the heads first.

I know Uber calls their service "Black". But in hot climates white cars make more sense. Burn less resources keeping cool.

Their cars are white in Tel Aviv (as all taxis, in fact).

Is the service there called UberBlack?

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact