Does Uber Know it Can’t Make Money Unless it Uses Self-Driving Cars and Gets Rid of Human Drivers?

Something is out of whack on the expense side.

Waymo, the self-driving car unit of Google’s parent company, filed a lawsuit on Thursday in federal court against Uber and its recently acquired startup Otto, accusing Uber and Otto’s founder Anthony Levandowski, who’d been a project leader at Google’s self-driving-cars project since 2009, of stealing a huge pile of trade secrets.

The New York Times:

In a federal court filing in San Francisco, Waymo said Anthony Levandowski, who runs Uber’s autonomous car division, downloaded 14,000 files from Google a month before leaving to start his own self-driving car company, Otto. Uber acquired Otto in August for $680 million, about seven months after Mr. Levandowski left Google.

“Otto and Uber have taken Waymo’s intellectual property so that they could avoid incurring the risk, time, and expense of independently developing their own technology,” the company said in the filing. “Ultimately, this calculated theft reportedly netted Otto employees over half a billion dollars and allowed Uber to revive a stalled program, all at Waymo’s expense.”

This lawsuit sheds new light on a problem Uber has, a problem that transcends all other problems, of which Uber has plenty: In 2015, Uber lost $2.2 billion; in 2016, the losses are said to have jumped to $3 billion.

Net revenue – the amount left over after it pays its drivers – was expected to exceed $5.5 billion for 2016. In the third quarter alone, it lost $800 million on $1.7 billion in net revenue.

But it’s even worse; these figures are based on sources cited by Bloomberg in late December:

Those are rough figures that may underestimate how much money Uber is losing and don’t include interest, taxes or stock-based compensation.

But it doesn’t seem to matter to investors who’ve been eager to throw vast sums of money at it. Uber has received a total of $12.9 billion in funding, according to the Wall Street Journal. The last round of funding valued the company at $68 billion, which is 37% higher than Ford’s market capitalization of $49.6 billion, though Ford generated $152 billion in revenues, a net income (GAAP) of $4.6 billion in 2016, and an “adjusted” (non-GAAP) income of $10.4 billion.

At Uber, the problem is a quirk in the equation: The more revenues grow, the more money it loses. Something is out of whack on the expense side. Or the business model doesn’t work.

“It’s hard to fathom Uber operating so far from profitability at a time when it feels like an established mainstream brand on the global stage,” TechCrunch mused when the losses were leaked. It pointed at the problem that has now become even hotter with Waymo’s suit against Uber:

Human drivers remain a serious cost center for Uber, even as the company has shifted its compensation practices over the years to lower the cost of every trip it makes.

Besides having to pay drivers, Uber has to fight competitors to keep them working  within the Uber marketplace, which takes incentives, bonuses, advertising and a solid driver-side mobile app.

Of course, Uber has also spent money defending itself in multiple lawsuits filed by drivers around employment classification and more.

Uber is spending enormous resources on self-driving-car technologies. This has unfolded in two ways: doing its own research and acquiring other companies, including Otto, the defendant in the Waymo lawsuit.

“There’s an urgency to our mission about being part of the future,” Uber CEO Travis Kalanick said about the acquisition of Otto at the time. “This is not a side project. This is existential for us.”

So if Uber could just get rid of its drivers and all the affiliated expenses and headaches? Self-driving cars would accomplish that. Uber made very aggressive moves in that direction, in its usual manner: throwing vast sums of money at it and ignoring the rules and laws that are in its way. The New York Times:

Late last year, Uber, in defiance of California state regulators, went ahead with a self-driving car experiment on the streets of San Francisco under the leadership of Anthony Levandowski [the former Google guy at the heart of the lawsuit]….

The experiment quickly ran into problems. In one case, an autonomous Volvo zoomed through a red light on a busy street in front of the city’s Museum of Modern Art.

Which was caught on camera. Luckily, no one was hurt or killed. Uber claimed it was human error. The New York Times:

But even though Uber said it had suspended an employee riding in the Volvo, the self-driving car was, in fact, driving itself when it barreled through the red light, according to two Uber employees, who spoke on the condition of anonymity because they signed nondisclosure agreements with the company, and internal Uber documents viewed by The New York Times.

All told, the mapping programs used by Uber’s cars failed to recognize six traffic lights in the San Francisco area. “In this case, the car went through a red light,” the documents said.

The description of the traffic violation reflects Uber’s aggressiveness in its efforts around self-driving cars….

But why is Uber so desperately eager to get self-driving-car technologies off the ground? It’s not an automaker; it’s not going to build and sell these cars.

It all comes down to drivers. They’re costly. They get paid for time and mileage. And as TechCrunch pointed out, there are the affiliated expenses of maintaining and increasing this army of drivers and keeping them from working at other rideshare companies. This includes the costs of recruiting, advertising, incentives, and bonuses. There are the management resources drivers consume, and the costs of “defending itself in multiple lawsuits filed by drivers around employment classification and more.”

These are variable costs; they increase as revenues increase. They crush the benefits of economies of scale. They might in the end reduce the business model to a no-profit or low-profit enterprise in a fiercely competitive world. For Uber and its investors alike, the way forward would be the hope of ultimately getting rid of the drivers – and all the affiliated expenses. Hence, the utter desperation with which it pursues its foothold in autonomous-vehicle technologies.

With this much money being thrown at autonomous-vehicle technologies by Silicon Valley companies and automakers alike, the industry appears certain they’re happening. But we’re not prepared. Read…  Self-Driving Vehicle Revolution to Wipe Out 4 Million Jobs

Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the beer and iced-tea mug to find out how:

Would you like to be notified via email when WOLF STREET publishes a new article? Sign up here.

  111 comments for “Does Uber Know it Can’t Make Money Unless it Uses Self-Driving Cars and Gets Rid of Human Drivers?

  1. RepubAnon says:

    Uber’s model seems to be selling their services at a loss to put their competitors (licensed taxis) out of business. Then, the prices will jump up.

    I’m not sure how self-driving cars help them – double-especially building their own. After the large R&D effort, they’d need to buy and maintain a whole fleet of cars. Plus, I’m betting that someone will figure out how to pull a Luddite-type monkey-wrench into the safe operation of such vehicles. Once that happens, the glamour of self-driving cars will vanish.

    • chris Hauser says:

      spot on. i can see a scenario of playing existential chicken with a self driving car, a la captain kirk, and the self driving car exploding in a shower of sparks and smoke.

      on the other hand, i can see a scenario of sale leaseback self driving cars operating from my automated robot staffed parking/repair facility.

      one of my all time favorites, and a formative influence:

      • Sonomendo Steve says:

        HaHa! I looked at those ads and then imagined “easy self driving car mod plans $10” in the back of Popular Science… to the 200mpg carbs.

    • d says:

      “I’m betting that someone will figure out how to pull a Luddite-type monkey-wrench into the safe operation of such vehicles. Once that happens, the glamour of self-driving cars will vanish.”

      Autonomous vehicles, mixed with human driven vehicles, in volume,in traffic, are far from an approved happening thing.

      Yet Uber seems to think it is going to get them approved for public transport, putting millions of drivers out of work world wide.Very soon.

      In this case the “Luddite-type monkey-wrench” is a simple thing called, reality, and another called, public safety.

      2040/50 maybe, 2030 no show, and can Uber afford to keep,on losing money they way it is, till the late 2030’s.

      With its current, obtain market share buy undercutting the competition, at an unworkable price, business model.

      Ubers current Business model, and fee structure, is an unfair trade practice. Their own numbers prove that.

      Some young lawyer, who wants to make a name, and knows a risk taking speculator, could look at that.

      • RepubAnon says:

        I’m thinking that a bunch of GPS jammers and some disco balls hung at random spots would throw a serious challenge to a self-driving car. Imagine some disgruntled activists (or agents of a foreign country) deciding to do this at busy intersections after self-driving cars became widespread.

        The current logic reminds me of the folks who originally designed e-mail and the Internet – the thought of bad actors never entered their thinking until afterwards, when SPAM and viruses started showing up.

        • david says:

          Many things from hacking to maybe even a large bug going splat on the optical sensor blinding it. Or it icing over, going snow blind, etc. All it takes is just one major accident and the insurance rates alone will kill it. Other than maybe a ride around the block out of curiosity I would never ride in those things.

        • Wolf Richter says:

          There were about 40,000 deaths in the US in 2016 due to human-driven vehicles, up 6% from 2015. Humans are notoriously terrible drivers.

          Now that humans are busy with their smartphones while driving, they’re getting more and more terrible as drivers. Hence the sharp increase.

          Self-driving vehicles only have to be better than these terrible human drivers – and that’s easy. They don’t have to be perfect.

        • Dan Romig says:

          Yes indeed, people texting while driving is maddening to me. As a tall person who bikes and motorbikes in MPLS/St. Paul, I can look down on drivers who’re concentrating on their phones instead of the road. The percentage of cars and trucks with idiots behind the wheel texting is insane.

          Driving is an active not passive endeavor, eh?

        • RF says:

          I don’t understand the argument that terrorists, pranksters, and bored geeks will cause all kinds of mischief and mayhem for autocars.

          “A combustion engine!?!? Why, that might as well be a bomb on wheels. What would it take to weaponize that 2,000 pound hunk of metal and kill and maim innocent pedestrians.”

          “Once we build all of these highway and rail overpasses there will just be a bunch of kids constantly throwing rocks off the bridges onto the cars for fun. It will be pure insanity!”

          Shall I continue?

        • Your Mom says:

          RepubAnon: so you’re saying the internet and email shouldn’t have been invented as quickly because….spam?

          I’ve heard lots of arguments against self-driving cars, but this is by far the worst.

        • rdblakeslee says:

          re Dan Romeg’s: “Driving is an active not passive endeavor …”

          Can’t tell you how many times my wife’s got mad at me for telling her to stop talking when I’m in heavy traffic …

        • Smingles says:

          “I’m thinking that a bunch of GPS jammers and some disco balls hung at random spots would throw a serious challenge to a self-driving car.”

          Yeah… probably not. Billions and billions of dollars and some of the smartest engineers in the world working on autonomous vehicles and you think “disco balls hung at random spots” are going to be their downfall?

          Like, thousands of engineers will collectively smack their heads when these disco ball accidents happen and say “of course! how could we not think of disco balls!?

        • Valuationguy says:

          I highly disagree that self-driving cars ONLY have to be safer than human-driven ones.

          While intellectually that may be true…you have to overcome societal inertia and fears. Thus the ‘more safe’ requirement becomes MUCH more safe. Then you have to iron out the insurance requirements….who pays in the case of an accident? The manufacturer of the self-navigating system…the human ‘monitoring’ the car in motion, the navigation map provider, etc.

          Airplanes still don’t take off or land by autopilot (even in poor ambient conditions) and the ONLY reason autopilot is allowed is when there are no other plane in proximity. Airspace is highly regulated. Is ground traffic control systems going to need to be developed considering the extremely close proximity cars operate near each other?

          Once the self-driving car ARE a reality….what are REGULATORS going to demand be added. Redundant control systems, monitoring systems, etc. which aren’t initially envisioned in the cost equation because they are too expensive…

          Will the roads themselves need to be upgraded? Potholes…deep puddles (water planing)….heavy rain….dirt/sand/mud/etc.

          Seems to me we are (at least) a decade away…..and Uber is losing billions per year? Will investors allow another Amazon…or when they lose patient in light of the likely crazy economic upheavals likely coming via the debt markets over the next 10 years?

        • david says:

          It will also be perception. The first big accident will be plastered on the news cycle for days on end.

      • Julian says:

        I’m not sure why many countries around the world would ever allow self-driving cars/computers on their roads.

        How could they allow a whole bunch of potential car bombs to be driving around their streets remotely controlled from Langley??

        It just won’t happen – and plenty of other countries in that boat.

        It’s a massive security risk.

        I’ll also add – self driving vehicles are a great boon for terrorists who don’t fancy blowing themselves up.

      • MarkB says:


        Most vision recognition works by edge detection. Try and find the edge of a plowed road… especially what they call ‘snow removal’ here in Portland OR.

        • d says:

          Or the edge of “roads ?” in western Australia, or the Northern territory, that are marked by softer sand.

          Just like much of north africa, and the greater ME.

          Developing a self driving vehicle, is not a have/con.

          Having it in the wild, mixed with human driven/controlled vehicles, on public roads, in urban traffic, at least 20 years plus.

          Now, we have sections, of separate walled off bus lanes. That network, could easily be extended and automated.

          That however does not help uber, as the working mining company tech, can be easily adapted, to such a network. The only real difference being the cargo, and how it is loaded, and unloaded.

    • Mike Levine says:

      Leaving aside the notion of an autonomous vehicle trying to navigate The theater district at 7pm, this article points out the folly of the whole notion of “ride sharing”. These companies initially positioned themselves as technology to connect people who want to take a passenger along on their daily ride in their personal vehicle with potential riders. It was clear to everyone except the politicians eager for lobbying dollars that this was never going to be the case. The drive (pun intended) to develop self driving vehicles is the proof that Uber, from early on in its development, has once again lied to politicians, the public and its “driver partners” about its goals and intentions. Upside? They can use the autonomous vehicles to ferry their attorneys and HR people to court when the sexual misconduct suits start rolling in. As long as the court is in a relatively traffic free zone.

  2. greg says:

    Cities and towns are now forcing Uber to buy insurance and licenses from the city or town where they operate. There are probably other fees and costs that gvts are now extracting from them too!

  3. RE: At Uber, the problem is a quirk in the equation: The more revenues grow, the more money it loses.
    As Uber is a transnational corporation, a more likely explication is tax evasion through transfer pricing, with the profits accumulating tax free in one or more tax havens. This is yet another anecdotal argument for unitary corporate taxation and a gross corporate receipts tax. A return to “normal” interest rates will almost certainly result in the immediate insolvency and bankruptcy of Uber and the other “Unicorns.”

    • chris Hauser says:

      the licenses and patents being held offshore, of course.

      too bad i can’t send my brain to the caymans.

      or can i…….actually, no.

  4. mean chicken says:

    Robots don’t earn $15/ or have health insurance expense. Does Bart still smell like a zoo?

  5. akiddy111 says:

    Slightly o/t :

    Do we think that truck driving jobs will start to disappear within 5 years as some commentators are forecasting ? If so, that could be disruptive as there are 4 million of them.

  6. Sonomendo Steve says:

    I’m pretty tech and auto-tech savvy, and the reasoning behind this “self driving car craze” (for lack of a better phrase) baffles me. I doubt we will see these common for at least 20 years, if EVER, maybe on special fast lane “tracks”. Kinda like fusion utility plants, ya know?…many decades at that one and still going.
    Do tech types like myself have some say in starting the whole effort, OTHER than getting to play around and make good money, which DOES make total sense to me? It is well paid FUN!
    We now have throttle-by-wire everywhere (which likely cost some lives; the Toyota glitch), but other than slight adjustments by computer, we do NOT yet even have steer-by-wire or brake-by-wire.
    The problem is the severe problems default. In throttle-by-wire it is simply a SPRING in the throttle body, which you all know as the “limp-home-mode”.

    Oh well, let the visionaries have their media and their glory, we just love tinkering for pay.

    • d says:

      Throttle by wire has a bigger problem, add salt water, sit back and watch.

      The amount of them being replaced by morse cables, in the marine environment, after issues, is massive.

      With marine throttle by wire, its WHEN there is an issue, not if.

      Self drive cars in the coastal environment, fatalities, waiting to occur.

      • Sonomendo Steve says:

        Yep! 99% of problems on most all “modern” equipment are electrical connections. Going to NASA standards is very expensive. More and more things moving into the “throw away” category every day. Fits the best of business models….RENTAL.

      • DH says:

        Passenger cars have already had throttle by wire since the 90s, so I’m not sure that’s a big issue.

    • Suzie Alcatrez says:

      The new dump trucks being built for mine work by Komatsu do not even have a place for a human driver.

      • Norman says:

        Off road, in private properties, with predefined routes, remotely monitored with a 100% system override capacity if something goes awry. And that´s notwithstanding the fact these dumpsters are the size of two condos, so you have the room of a nice kitchen to place redundant specialty electronics to run the whole thing. Not remotely comparable to a passenger car. If 100% automatic functioning was practically feasible, the first vehicles operating this way, apart from military stuff, will be commercial airplanes, not even trains are remotely close to that. (Yes, I know about Singapur´s robo-underground. It´s not a train, the tube is nothing like a real passenger or cargo train, by a very long shot) Yes, there´s a lot of research going on, many electronic assistance deployed and some apparently astonishing capacities, making many people believe the goal is closer than it really is. The truth of the matter is the hard parts are barely beginning to be addressed, there´s a ton of extremely complex problems to deal with, and don´t forget the final price tag has to come at some tens of thousands! Regarding human drivers, the main problem is poor education at driving schools, they are tought how to pass the driving tests, not how to drive properly. With a few extra driving courses, the vast majority become very proficient drivers.


        • d says:

          Another think with those fly by wire mining systems. They only go in big mines that can afford/justify the expense.

          Not all the new Kommie dumpsters, are fly by wire.

    • Graham P. says:

      Nothing baffling about it. The average American car sits idle for 95% of it’s useful life. If you were running a business, would you buy a depreciating asset that you used 5% of the time? Perhaps if there was no other option – but this technology is really quite close. Anyone who believes otherwise is fooling themselves. I don’t think it will take more than 5 years for self driving cars to become ubiquitous in urban areas – suburban areas will certainly adopt later, the countryside perhaps never.

      • Kent says:

        There is a reason for that idleness. 95% of use is commuting to work. Making them driverless doesn’t change that dynamic.

        • Smingles says:

          Sitting in traffic is considered “in use,” not idle. The 95% figure really means PARKED.

          Think of it this way… if you spend on average 90 minutes a day driving– regardless of whether it’s at 75 mph or 7.5 mph, your car is not being used approximately 94% of the day (22.5 hours)– which is just one of the ways of calculating the number, but they generally all come out to roughly 95% on average.

          Now, how that applies to Uber is a bit more complicated because of peak usage times. Sure, it’d be great to not have any spare cars just sitting around at any time, but you need to have enough cars to drive everyone around at peak times… which will then probably necessitate having idle cars at non-peak times.

      • Intosh says:

        “Quite close” IS exactly why it will not be widely used for another 10 years.

        Take speech recognition for instance. It has been “close” for at least the last 10-15 years. The technology can recognize human speech with 90-95% accuracy but it is the last 5% that has been a major roadblock for widespread adoption. That last few percentage is the most difficult to overcome both technology-wise and usability-wise.

        Now take self-driving, which may entail severe consequences to the user if the tech is wrong, and think how there could possibly be a market for it with such error rate. Even with an error rate of 1%, nobody will use it.

        And it is not even a question of probability; it is a question of fear. People are more affraid of terrorism than of guns or of driving, even though there is infinitely more chance of dying on the road or from gun violence. This is the psychology of control (or lack thereof) and uncertainty. So even with an error rate of 0.1%, it will be a tough sell.

    • Intosh says:

      Totally agree. Not gonna happen before a decade or two.

      This just shows, once again, Silicon Valley’s arrogance and hubris totally getting out of control.

      For Uber, they underestimated the workforce’s inevitable rising demands and the governments resistance of disruption. They dug themselves in a hole. Their bet on self-driving car is a losing bet — they will run out of time, even in this era of money printing. But I guess nobody at Uber wants to wake the king’s up from his fantasy dream because engineers there are handsomely paid to keep the fantasy going.

      • And then again “The King” may be well aware of what is happening, and is staying up late packing his bags, getting suitable (fake) travel documents, and moving his money to off-shore tax and money laundering havens. We are overdue for another Enron.

        • Intosh says:

          It will be like Theranos but 100 times bigger. Scam artist Elizabeth Holmes probably has a few million $$$ stashed somewhere while her start-up is crumbling. It’s robbery in broad daylight.

  7. Richard Hill says:

    The future might be a hybrid between self-driving and remote driving. The actual successful work with driverless mine trucks in Western Australia works like this:. The individual trucks can handle most tasks, but they are commanded by remote control from a control centre thousands of km away in Perth. One driver in Perth controls several mine trucks. Perhaps Uber can build a centre like Nellis AFB. Can you imagine a vast hangar with rows of Uber drivers controlling cars all around the world. The Uber cars do on their own the things that are easy to automate like driving on a freeway, and the remote driver does the rest.

    • d says:

      “One driver in Perth controls several mine trucks.”

      In those applications, if there is an issue, they just hit the e stop. Causes no problem’s, as all the truck movements, are Auto/by wire.

      lets see you “e stop” mixed auto and human controlled, 6 pm la traffic.

    • Sonomendo Steve says:

      Richard- The Post Office actually did that. Before the OCRs got good enough to read all the mail, at 7-10 letters/sec, pictures of unreadable letters were sent over T-1 lines to various centers with hundreds of computer operators reading what the local computers couldn’t.
      They were real sweatshops, the only regular Postal employees there were the few techs and mgrs, but they are all gone now.
      Computer power.

    • 30000footview says:

      One only has to look at the MILITARY DRONE operations. Nevada site controls drone operations all over the world. A single operator is able to control single/multiple operations depending on mission. Mostly fully licensed pilots at the controls…for now. The future of airline aircraft operation will eventually roll the same way. Sophisticated A/C controls are fully capable to take off, fly and land in automation mode. Pilots/co-pilots are basically very highly trained…but often times just system babysitters. Hours of flight simulator training….exactly duplicate every possible scenario …… Kf a simulator have replicate any flight situation….can it also completely replace a human operator? Cargo planes like Fedex and UPS will be the first to convert. The FFA will be slow to transition to to automated passenger plane operation. Will there be accidents…you bet….but nothing is 100%…especially any current system. its always rish analysis. Airline Pilots make upwards of $200-300K per year….United alone I think has 6000 pilots…u do the math?

  8. AC says:

    When Uber’s Self-Crashing cars self-crash into other people, who will bear liability?

    The passenger? Probably not. Maybe in New York City.
    Uber? Maybe.
    The car manufacturer?
    The self-crashing software subcontractor(s)?
    The government, for allowing self-crashing cars?
    An assetless Cayman shell corporation?
    Russian Hackers?

    Should be entertaining.

    • Wolf Richter says:

      The insurance industry is already all over this.

      • MarkB says:

        Volvo may set the trend. They are stating they would be responsible for the vehicle when running in self drive model

        • d says:

          Volvo is now owned by the chinese.

          They value the average western life at around $ 5.00 per head, and that’s all they might pay, in the event of a fatality.

          An admission of liability, from an insolvent corporation, that wants the case heard in its politically controlled courts, is worth nothing.

        • Wolf Richter says:

          If Volvo agrees to do this, it already has an insurance company lined up that will cover it. That’s what insurance companies do to make a living. And Volvo will pay for this coverage via premiums. There will be liability limits, and settlements will stay within those limits. This so routine in every other sector. It’s just a matter of getting the quirks worked out and figuring out how to price that risk.

  9. Petunia says:

    Self driving cars are at least a generation away. Uber concentrating on the software is a total waste of time because the real problem is the quality of the sensors. If we still can’t make cars that can tell us reliably, which tire is losing pressure or how much oil we have, the rest is wishful thinking.

    Then, there’s the problem of most techies knowing tech real well, and nothing about anything else, especially business.

    Then, there’s the problem they have with how they treat women employees. Not good business to abuse the female staff. Ask Bear Stearns how that worked out for them.

    • Mike says:

      A generation away? That’s 20 plus years! Have you ever heard of Moore’s Law?

      • Petunia says:

        Yes, I have, but this technology does not exist in a tech vacuum, therefore the generalization does not apply.

      • Suzie Alcatrez says:

        We will have basic artificial intelligence in 20 years.

        The future is getting quicker.

        • Petunia says:

          I’ve been hearing that since the 70’s.

        • NotSoSure says:

          The current so called “AI” is old technology dressed up in cooler monikers.

          The so called “Deep Learning” is really the same as Neural Networks from back in the 80s. SVM or Support Vector Machines is from the 60s.

          So what’s different today? Faster computers, and the ability to network them together to train the computer models.

          Still if you make an AI expert drunk, he/she will not be able to tell you why Bayesian statistics is proven to make all of this work. After all Bayesian statistics does not know about truth/facts, etc, the foundation of all reasoning (intelligence). What’s true about computing is still true about AI i.e. Garbage In Garbage Out. If you put in bad data into the most powerful model in the world, it will still spit out bad outcomes. That’s why Google always says that the data is more important than anything. The trick to fool all AIs then is to simply keep changing the inputs. By the way I was talking mostly about General AI. Computers can beat a human anytime when it comes to highly constrained environments such as games. Bu then again there’s no intelligence, it’s just brute force (computation to try out all branches) and statistical recognition.

        • RD Blakeslee says:

          re Petunia’s “I’ve been hearing that since the 70’s.”

          It took 200 years for the Luddites to be vindicated.

      • Tyler says:

        Moore’s law became irrelevant in 2012, at the latest.

    • Thomas H Belstler says:

      Vaclav Smil who has studied the acceptance of new technology for decades says that even if a technology is well accepted, it takes 20 years for it to diffuse through the economy. To me it looks like it is being force fed into the system without adequate development and testing in order to save money. Let’s wait and see what juries and courts have to say after the first 500 accidents at a minimum have been processed by the system.

      A single case wherein a verdict that the programmer should have foreseen the circumstances of the accident is upheld throughout the appeals process is likely to put a serious crimp in the process.

  10. manny says:

    Self driving cars are vastly, vastly safer than human, just like yaw controls. the camera and its processors analyses hundreds of millions of data bits. to date no study prove otherwise.

    • david says:

      what happens when that camera get hit/blinded by birds, bugs, rocks, ice, snow blinded, heavy rain, etc? what does it do when the camera can’t “see”?

      • manny says:

        multiple cameras, newer system has thermal, night vision, fail safe mode, if all fail, it pull over and park itself. like I said much safer to drive than humans.

        • david says:

          and if it is in the middle lane of a 5 lane highway in rush hour and a sensor craps out? when we have to throw on a blinker to pray someone lets us over? would be interesting to know when the sensors detect cars on every side and needs to get over….does it put on a blinker and wait until it senses a space?

      • NotSoSure says:

        Same as a human being i.e. driver is impaired. The point of a self driving car is not to eliminate all accidents, rather it will make some mistakes that a reasonable human being can make as well depending on the power of the sensors, etc, just not as often since a machine does not get tired.

        And for the current version of self driving cars, I think most will stop if the camera is damaged.

        • Sonomendo Steve says:

          I have tremendous respect for computer/sensor speed and integration. Compare 70’s NMR printouts to an MRI scan. Wow!
          But the inputs and situational awareness to drive a car? I still think that is a very tall order.
          But I am jealous of the guys who get to play with it.

    • d says:

      “Self driving cars are vastly, vastly safer than human,”

      Yes but they are not reliable, and they do not mix well with unpredictable illogical human’s.

    • Intosh says:

      It may well be but the psychology of fear, lack of control and uncertainty may throw any probability numbers out the window.

      A survey showed that terrorism is the second thing Americans are most afraid of, even though being a victim of one is much much less likely than being a victim of a fatal car crash or of gun violence.

      So it doesn’t matter that self driving is “vastly vastly safer” than human drivers — it needs to be damn near perfect and we are not there yet.

  11. Peter Kleinman says:

    The “self-driving” car operators need to have the municipalities reconfigure their streets for the convenience of these vehicles; possibly outlawing pedestrians crossing intersections.

    • MC says:

      I wouldn’t mind outlawing pedestrians who jump right under your wheels, a depressingly common feature of most urban areas

  12. Bruce Adlam says:

    Self driving cars are more about fleecing investors of more money .on public transport the wayfarer tag on tag off does not work 100% of the time and even with latest ones they are no more reliable they were 5 years ago. That will be the same problem with self driving vehicles. Try telling loved ones that a malfunction machine killed your kid. Will not be acceptable or tolerable

  13. fizzy says:

    The industry is way ahead of itself thinking autonomous cars are close to being ready for prime time. They’re already performing well in closed circuit courses, like mining sites. But integrating self-driving cars with human drivers, bicyclists, and pedestrians on public roads is a magnitude harder problem when the results have to be near perfect.

  14. Maximus Minimus says:

    I can see self-driving technology as a niche, and as a enhancer of driving safety. Some of the passenger cars already implement some features, e.g. distance warning when you have a GPS with camera.
    Long distance truck driving is a tedious and accident prone profession. A self-driving truck in non-urban areas is a good idea.
    Otherwise, this flurry of news is media blitz, and hopefully will die down as soon as there is some other popular (harmless) subject to cover.

    • RD Blakeslee says:

      Possible niche?

      Older drivers who might have to stop autonomous driving.

      I’m 85 – certified OK to drive until 110 …

  15. Paulo says:

    I’m pretty sure we will be hunting self-driving cars where I live. As they head for a pax pickup they will be run off the road. Of course in my locale people also shoot drones on sight. We don’t like jet skis, either…or rap, cell phones, …….. you get the picture. :-)

    “Make the world go away
    Get it off my shoulder
    Say the things we used to say
    And make the world, make it go away “

  16. Kevin Beck says:

    Similar to what happened in San Francisco, in Pittsburgh, Uber lost their relationship with the city because of excessive demands for special attention regarding city services. They wanted permission to use the bike lanes for parking, and also insisted that the city plow the streets that they used for their routes on a different schedule than the city already had. There were other details about this, but they just wore out their welcome with the city and Carnegie-Mellon University.

    • MC says:

      The big mistake Uber has done was going after big cities such as San Francisco and Pittsburgh instead of the small towns whose council members are seemingly obsessed with ending up in the news.
      There are thousands of such towns in both the US and Europe that would have gladly thrown their citizens/taxpayers under the bus to accomodate Uber for a big title.

      That Simpsons episode about the monorail was more documentary than fiction…

  17. Nik says:

    Aloha friends…the Major Economic Disruption everyone over Looks with Self-driving Cars,as they do with Robots…Neither…Will SPEND nor CONSUME in the marketplace..! thanks for reading,aloha

  18. NotSoSure says:

    I would suggest people to Google Naked Capitalism Hubert Horan. He is a transportation expert and he did a super detailed analysis of Uber’s business model over on that blog using known numbers. Truly required reading if you want a deep dive analysis of the business.

    TL;DR, Uber will never work.

    Here’s another good article:

    Everything that Hubert Horan pointed out like driver subsidies is turning out to be … true, just witness the Uber + Ola strikes in India.

    Self driving cars will only work given current infrastructure if they can navigate the streets in 3rd world countries. Remember the infrastructure in the US is not getting better.

  19. Redman says:

    Self driving cars will monetize every aspect of their operation.

    Uber of Ford or Honda will record the passengers lives in and out of the driveway; in addition to their neighbors properties and family life. (Imagine an Uber camera with the necessary night vision focused on your driveway from across the street when it’s parked overnight).

    And you can be sure law enforcement will want instant access to both gps location as well as the camera function because, uhh public safety. (Many shop owners around and in NYC can tell you they are forced to install police accessible video feeds!)

    Since the public is too dumb to drive, this implies they are too dumb to live without the everpresent intrusion of the civil authorities.

    But the biggest downside to these cars will be the fact your boss will want you to work in the car!

    • Kent says:

      And I expect a big screen in front of you blasting advertising for the
      entire trip

  20. Mike R says:

    The demise of self driving cars from ever getting off the ground ? One word: snow.

    • Wolf Richter says:

      Have you ever seen humans drive in the snow? Or worse, in a blizzard? LOL. Humans are blind in snow. They’re stupid in snow. They’re clumsy. They’re just terrible. Just check the TV coverage when there are 3 inches of snow on the highways in Atlanta – all human-driven vehicles….

      • Dan Romig says:

        I, and those of us Minnesotans who actually switch to snow tires (Dunlop Winter Maxx or Bridgestone Blizzak) take exception to that generalization Wolf. In fact, the smart way is to use narrower tires for snow and wider for performance. My car has 225/55/16 as the stock size, but I go with 235/45/17 in spring, summer & fall. For winter, 215/60/16 works the best.

        OK, I will admit that I’m a detail-oriented motor head, but I am not alone. Quite a few other Minnesotans have all-wheel drive (RWD for me) performance cars equipped with snows for winter.

        • allan says:

          As a Canadian very familiar with driving in snow, I agree with Dan although Wolf’s point is still valid for those people who don’t drive in snow frequently.

      • RD Blakeslee says:

        Above the arctic circle they, and above 3000 feet in WV we, use dogsleds …

      • Maximus Minimus says:

        But humans can learn from their mistakes, can software instantly learn from a near miss and apply correction to future action. The computer would have to beat maybe 80 % of human drivers to make driving safety better.

  21. PanamaBob says:

    At almost 68 years, I’ll comment on Moore’s Law. It is doubling at a period of 18-24 months on average. I see technology along with automation doubling at a progressively shorter period, maybe soon approaching 12 months and less farther out. Things are really moving fast with major breakthroughs in science like genetic knowledge and manipulation, search CRISPR.
    I wish it was slowed down but it is what it is.
    Alas, where will the future jobs be for my grandchildren?

  22. John k says:

    Humans cause lots of accidents. But uber accidents might generate larger jury awards. And competing taxi drivers might figure out ways to attack self driving competitors.

    • Mark says:

      I think litigation risk aspect of this is way overblown. Yes, people will be injured and suits will be filed, but Uber will contract with an insurance carrier with all the resources to deal with these inevitable cases. A settlement will be reached within the limits of Uber’s liability coverage, rinse and repeat, rinse and repeat. Everyone makes a few dollars and Uber pays a bit more for the cost of insuring their business. This is just a another cost of doing business.

  23. opalchip says:

    There’s two discussions going on here – –
    1. When will self-driving cars be ready for prime time?
    2. Will Uber make it that far?
    IMO the answer to number 1. is, “Who knows…” There’s lots of factors, not the least of which is politics and regulatory. What if Colorado doesn’t want them and Wyoming doesn’t? Can the feds force them down every state’s throat, even on local roads, as interstate commerce? Who services tens of millions of them – reliably? What sort of ID will be required (in the future) for you to “hail” one, and which databases can the ID be run against – while you’re still in the car? In fact, they won’t even need ID, there will be facial recognition in the car. If you’re undocumented – will the car redirect your trip straight to the border :) ? All sorts of stuff to resolve/build out that will take years – compared to running a successful test with a few hundred vehicles in a local area.
    Number 2. – stranger things have happened than Uber survivng, like AMZN going so many years without profits just by using it’s ever-inflating stock value as real currency, but I think it unlikely in Uber’s case. At the point where Uber (if they get clear of these lawsuits) has access to affordable, functional and legal self-driving cars – so will everyone on the planet! The barrier to entry to compete with Uber will be low for local companies who know and market to their own little area. If I owned a self-driving car and wanted to timeshare it out when I don’t need it, I would certainly prefer a local company with an office that I can walk into and deal with – especially in the case of a problem. I would certainly be willing accept a somewhat smaller payout for that convenience, especially knowing that the profit/income was staying local.

  24. ian says:

    Not sure I understand the self driving thing. Their business model, notwithstanding the fact that they are losing money, is based on bringing customers and suppliers together and taking a cut of the rent. They do not own any vehicles, they do not have driver employees (yet), they do not have vast maintenance or vehicle storage costs. The model was a new one and therefore had first mover advantage although that is now under threat by others. If they go for driverless cars they must own the fleets surely with all the expense that incurs and everyone else is aiming for the same thing so no advantage. What am I missing?

  25. Kent says:

    Uber has no advantage that isn’t easily replicated. It’s an app and not a terribly sophisticated one. My guess is 5 years from now it will go down as one of the world’s great scams.

    • MC says:

      Uber’s model is similar to Tesla’s: it works as long as both aging hipster fanboys and yield-desperate investors are willing to sink money in their financial black holes.
      How long will it last? Who knows, but I am willing to bet either until a Theranos-level scandal or yields on traditionally safe assets going back to pre-2009 levels.

      • DH says:

        Which hipster fanboys are you talking about when referring to billions of dollars of investment? If anything, my more liberal friends are turning on the business practices of Uber. Sure, it’s easy to out perform a taxi company, when you side skirt regulation.

        Ultimately, we’re just seeing investors throwing money at everything to see what sticks.

    • Petunia says:

      I could never understand what the taxi fleet owners were complaining about. Hire two programmers to write and maintain an app. Problem solved.

      • From an immediate operational perspective you are indeed correct, but from a total business perspective Uber is simply avoiding/evading most of the business expenses of the cab companies.

        One major expense is the franchise tax many cab companies must pay to operate in a municipal or service [e. g. airports] area and in many markets, the cab medallion which authorizes an individual cab to operate, which in many cases costs tens of thousands of dollars and is more valuable than the vehicle, e. g. NYCes. In many cases the cab companies own and maintain their vehicles, incurring property tax on these assets, and cab drivers are required to have special licenses which require criminal background checks in addition to chauffeur/CDL state licenses which require periodic random drug testing.

        While there is an element of “rent seeking” in the legacy cab companies regulatory costs, these regulations evolved for a reason such as safety, and minimization of fare fraud/extortion. Uber avoids [or more exactly “externalizes”] these costs, but also avoids the benefits to the public produced by these regulations.

        The effort by Uber to introduce “self driving” vehicles appears to largely be an accounting mind game in that the actual cost savings/avoidance will be minimal, although it will shifted between accounts, transforming what is currently a variable cost/expense to a marginally lower fixed expense, but with a vastly larger depreciating asset base (the vehicles) which must be owned by someone (and how do you lay-off an autonomous vehicle when business is slow?)

        Perhaps the major benefit will be the much better street/address/route knowledge of the autonomous vehicles which will be able to optimize the routes for cost, speed, etc. depending on current road conditions, time of day, etc. FWIW – a fully automated vehicle seems like “overkill” for this in that a computer algorithm at Uber central combined with a GPS system in the driver controlled Uber vehicle can do the same thing at much less total cost.

        As noted by other posters, Uber may well be the 21st century analog of the Tulip mania which was making everyone in 17th Holland century rich, or the railroad speculation mania that swept England in the 1840s.

      • Ross says:

        Bingo Petunia,
        40 yrs ago I drove cab in Toronto. An app would tell you who you were picking up, where they were going, and your order would be computed by a unit that allocated your business in an unfavored way to make it fair and useful for all. And of course the customer would know who was picking them or their packages or loved ones up and that they would also be treated fairly and efficiently in unit allocation. Who needs Uber then?

  26. opalchip says:

    When self-driving cars are safe enough to be ubiquitous, there will be an infinite amount of competition, pushing prices to very low margins. When they are THAT safe, insurance companies will off 100% no-fault (for everyone involved). There will be winners in that scenario, but it’s hard to say who they will be.
    I think it will work this way – When you know you don’t need your car for some period of time – 2 hours or 2 days – say, while you’re at work M-F, you will log into a carshare app, toggle your car as available, specify when you need it back, and log out. The system will know where it is and if someone nearby orders a car, and yours fits the bill (and can get back to you in time) it will take off on its own, return, and there will be a $$$ credit in your account. There will be many owners willing to offer their car at BELOW rates that would allow a for-profit dedicated taxi service to make money – because the individual owner just wants enough to help pay off the car loan.
    A variation of this, btw, is already going on in China, where people buy higher end cars that are popular rentals for wedding caravans, which they wouldn’t normally afford. They use them as daily drivers but rent them out on weekends to help pay the loans.

    • Kent says:

      Why do you need a self-driving car? Why not just have an Uber driver pick you up in the morning and drop you off at work. He can take fares for the rest of the day and pick you up at work and drop you off at home. No need for a car or loan in the first place.

      But this doesn’t work because of the same scenario you have above: the number of people needing rides to and from work are orders of scale greater than the number needing rides for the rest of the day. So it doesn’t scale.

  27. cdr says:

    Self driving cars- just around the corner. Call a self driving cab on your cell phone. An app will route it. No people needed for anything. A globalist’s dream.

    What is it about self driving cars that captures the fantasies of so many people? The convenient rationalization is ……

    * Big companies are spending lots on this
    * Accidents – the insurance industry will deal with it
    * Lawyers – don’t worry about it. The insurance industry is dealing with this. Lawyers don’t matter. Laws will prevent them from doing anything about the problems that most certainly WILL NOT happen.
    * Look!! Company G, or U, or XYZ just issued a press release about how valuable this technology is / will be. IT’S COMING!!!! Woe to the naysayers. 3/4/5 Years tops! Ha Ha negative nabobs. See.
    * Did you see the JonnyCar on that movie about Mars? And you doubt me about this?? Loser. Self Driving Cars are almost here.

    I’m waiting for the IPO for a ‘research company’ with a sure thing being offered, subject to one or two or dozens of qualifiers in the prospectus.

    • cdr says:

      Self Driving Cars – Evidence that people need to believe in something bigger than themselves.

      Hey buddy – want to buy some stock in a company working on a sure thing? Just look at these sales projections. Everybody uses cars. Just look out the window. Price: buy now because it’s rising fast. FOMO deserved here.

    • cdr says:

      Prediction: The first pure play self driving car research companies will have a substantially oversubscribed IPO. Price per share will be calculated as a multiple of anticipated burn rate of cash received. Capital gains will be whispered as a no-brainer and free money to the astute investor. An actual bargain.

      Evidence? Enthusiastic bloggers jump down the throats of anyone who offers even a little push back against their dreamland. It’s like Hollywood vs Trump. Sucker bait.

      • Sonomendo Steve says:

        You are right cdr. I once had a PhD in plasma physics from MIT tell me given enough time and money he could build a tomato juice laser. Lets hope that notion never becomes part of the beloved collective “dreamland” you refer to.
        For some reason PT Barnum comes to my mind.

        • Sonomendo Steve says:

          Since you mentioned Trump, let me make it clear I’m a Bernie guy. There are some “dreams” that are moral/existential imperatives. Science and tech are searches for physical truth.

  28. prepalaw says:

    Rolf – two comments:

    Otto. There has been a self-driving truck named Otto (on road and off-road – like in a warehouse) build in Kitchener Canada for many years.

    Why would Uber name its competing product Otto. Uber is being sued for trademark infringement, etc.

    I see Uber as a company with poor management and too much money.

    You say self-driving cars – “They don’t have to be perfect.”

    I think you are wrong. Imperfect machines can not be let loose anywhere running amok – especially on public roadways.

    Self-driving vehicles need at least a six sigma level of accuracy – that is 3.4 mistakes in 1 million decisions. And that many not be good enough for a driverless car. Look at commercial aircraft – the guidance system is pretty damn good – but you still have at least two human pilots in the plane.

  29. Kent says:

    So here’s my issue with self-driving cars: what do I gain? Certainly they will cost a lot more due to all the sensors and computing equipment. And for that extra money I don’t have to push on an accelerator with my right foot? I’m honestly not that lazy.

    I think it might be an idea for 80 year olds to get around. But then they may not have the extra money to buy one.

    And Uber. When I want to go somewhere, I don’t want to wait 20 minutes for some stranger to come pick me up and sit in a car where the last guy might’ve had the flu. I like to get in my car and go. Now. And at 80 mph if I want.

    I just don’t see these things as an improvement in my standard of living.

    It’s like 3D printers. Ever used one? Put in $20 worth of filament and tell it print a cup. 3 hours later you have a $20 plastic cup.

    • RD Blakeslee says:

      “I think it might be an idea for 80 year olds to get around. But then they may not have the extra money to buy one.’

      Will us 85-yearold need two of them, the second to be a “wingman”, sort of?

      Payment: Medicaid.

  30. Matt says:

    The evidence that the tech is ready for prime-time is the lawsuit itself. We can assume that Alphabet would not waste social and financial capital engaging in a gargantuan legal battle over questionable tech. Uber isn’t some little 5-person shared-space upstart with no legal team.

    That doesn’t mean Uber’s Hail Mary will work. If it looks, smells, tastes and feels like a tech theft, it’s probably a tech theft. But why would Uber bet the farm and blatantly steal from deep-pockets Alphabet?

    Uber’s current biz model is to shift wear & tear + maintenance onto its “contractors”. It competes against cab companies that figured out long ago how to find $500 ACV cars and wring a little more use out of them. The question is whether a self-driving car is cheaper per-trip than a fully depreciated, human-driven Crown Vic with 200k still left in it, allowing for the occasional junk-yard-part repair?

    Right now that answer is unknown. What *is* known is self-driving tech *will* replace long-haul truck drivers because self-drive tech represents a much smaller fraction of the overall vehicle value. There’s no such thing as a $500 ACV tractor that has any useful life to be squeezed out. There’s no personal tractor truck market through which they can be depreciated first.

  31. David in Texas says:

    From the original post: “Net revenue – the amount left over after it pays its drivers – was expected to exceed $5.5 billion for 2016. In the third quarter alone, it lost $800 million on $1.7 billion in net revenue.”

    Wolf, I’m just curious. If Uber is taking in $5.5 billion in net revenue after paying drivers, where is this money going if they are incurring such huge losses?

    It’s easy to get caught up in the argument as to when we will have fully autonomous vehicles, but it seems to me that the more relevant question for Uber is why it can take in $5.5 billion AFTER paying its drivers and still not make money. It doesn’t look like driver pay is the company’s main problem.

  32. JimTan says:

    I agree that getting autonomous cars to work soon (which is unlikely) may be the only way Uber survives. This resulting push into R&D spending for autonomous vehicles is probably contributing to their widening losses.

    Another expense likely contributing to their ballooning yearly losses are legal, lobbying, and PR costs. Uber’s designation of drivers as low cost/no benefits Contractors instead of Employees is running up against amny class action legal challenges. With over 600,000 drivers, some will eventually challenge the fact that they have no guarantee of minimum wage, health benefits, paid leave, while bearing personal responsibility for vehicle lease payments, gasoline, road tolls, car insurance, and parking. Uber has not even reached the point of serious regulatory challenges for labor law violations. I wonder what portion of their high operating expense is allocated for legal, lobbying, and PR costs to prevent regulatory challenges and mitigate class action settlements? There have been some indications over the last two years of Ubers rapid escalation of resources in these areas:

    Lawyers, lobbyists, and PR across multiple cities and countries are expensive.

  33. I think that the Uber model is limited in its utility to society:

    1- It increases marginally traffic – which increases congestion to a much greater extent.

    2- The drivers need to be paid – as this article made clear.

    3- Autonomous cars may solve the driver problem, but they will make the congestion problem much worse – empty cars wondering about waiting for passengers.

    To solve the above, I have developed a system to allow people to get lifts from others travelling the same way. You can try it out by clicking on “Simulator” at It is currently undergoing test and I hope to launch it in the near future. The software can be readily adapted to any other place covered by Google Maps. Here are its advantages

    1- The number of passengers per car increases – more people get to travel during rush hour while improving congestion.

    2- The technology is here today and it works – commuting by car works.

    3- The cost of the driver is trivial – since he/she is driving that way anyway.

    BTW, 3 people from Uber have checked my linkedin profile over the past week. I would like this service to be free and for passengers to pay a small amount to drivers – depending on distance travelled. Ideally, drivers and passengers should swap places to cancel out these payments.

  34. Gregg Armstrong says:

    Uber’s business only works because its intellectually challenged drivers provide their own automobiles and pay all vehicle related expenses. So, if Uber gets rid of its drivers who will pay for the capital investment in automobiles, maintenance, upkeep, fuel, parking and insurance costs? The robotic AI application? ROTFLMAO.

  35. Robert McGregor says:

    Authors like James Howard Kunstler have argued the regular human-driven car design itself is negative and counter-productive. It destroys the environment and human societies. It is too complicated and expensive for what you get! We would be better off with more light rail, buses, rapid transit, more walking, biking etc. Adding “auto-driving” to an already excessively complex system will be “two steps too far.” The auto-driving software may work fine “in isolation,” but the reality of imperfect roads, imperfect weather, financial and insurance complexity, and human sabotage will bring down the whole project. Not to mention the 4 million driver jobs which will be lost. If we were more enlightened as a society, we would simply stop this this project. It’s benefits cannot exceed its costs.

Comments are closed.