Uber sells self-driving unit in deal that will push Aurora’s valuation to $10B
techcrunch.comWasn't Uber's whole money losing strategy based on the vision that one day they'll use their self-driving cars to get the driving done for cheaper?
No, that was never the strategy, no matter how many times people on HN claimed it was. Both Uber and Lyft are aiming for profitability next year with human drivers.
(Disclaimer, I work for Lyft but this has been said in public many, many times by both companies.)
One of their execs literally pitched me this very vision when I turned Uber down for an eng mgr position several years ago.
I thought it was comical that his pitch was "The reason we're going to be successful at Problem 1 (taxis) is because we're going to be successful at ridiculously-difficult Problem 2 (self-driving) which we're not close to solving". Pass.
I love it when job interview pitches fall flat like that. Many years ago, I had a job offer from IBM for a class of server I wasn't familiar with. The guy explained it to me like this: "You see, desktops these days are running Linux, and mainframes are running (IBM OS something?). We represent a class of server bigger than desktops but smaller than mainframes. It's an exciting space. The desktop class of PCs are handling bigger and bigger workloads every day, and the mainframes are getting cheaper and smaller all the time, too...." and then he sort of trailed off and we sat quietly for a few seconds, and then he started talking about the nice office.
If my guess is correct, it seems that you have missed a chance to work on relatively popular AS/400 servers ... :-)
I have no idea, but the engineer I was talking to was describing the unique challenge of working with the OS, as it was the sole system compiled in that language, and how about half of the bugs they fixed were OS bugs and the other half were compiler bugs, and I ran away from there as fast as I could. This would've been somewhere around 2005, so it could've been.
I see. Yeah, it sounds pretty scary. Well, hopefully, the job that you have accepted instead of that one has been much better.
Could you explain why are AS/400 relatively popular? I don't read much about these mainframes, and am a user of Personal Computers (macbook, thinkpad) and EC2 instances (something that runs Linux?) , so I am not sure what AS/400 servers are, but I'd love to learn more!
Well, by "relatively popular" I meant that AS/400 servers (aka AS400 or IBM iSeries or IBM System i: https://en.wikipedia.org/wiki/IBM_System_i) have been quite popular mostly only within certain industries and/or niches (e.g., banking, insurance, manufacturing, supply chain). They are still widely used there for legacy reasons, however demand for AS/400 developers (which implies proficiency in RPG language, since, as I understand, most legacy applications are written in RPG: https://en.wikipedia.org/wiki/IBM_RPG) is tiny; so, it's a good language to know to potentially expect a relatively high salary or consulting rate (hello, Delphi!), but a dead end for IT career growth. I would also venture to guess that some (many? most?) companies currently using AS/400 plan to migrate their legacy applications to the cloud.
I wouldn't want to work on the AS/400 systems, but its Technology-Independent Machine Interface (TIMI) was basically Android Runtime decades before flip phones were even a thing: ship all of your userspace binaries as bytecode, and compile that bytecode to optimized machine code for your CPU at program installation time.
That's quite fascinating. I appreciate your comment.
"Popular" was a touch sarcastic. AS/400s were most recently called the IBM system I, and are now a discontinued product. IBM still sells mainframes, and they have discontinued the AS/400.
They are most certainly not a discontinued product. We just purchased a brand new Power9 server at work. The model we purchased was just released. Everyone at work still calls it an AS400. They are bulletproof and run many large businesses today. They are also not mainframes but are a class of machines called minicomputers.
IBM Power Series is closer to the RS6000, no? Running real AIX and having a full JCL makes them a far sight better than the RPGII stuff on the AS/400. Night and day.
IBM AS/400 is not discontinued, but rather has been rebranded as IBM Power System: https://www.ibm.com/it-infrastructure/power.
As a former IBMer you missed nothing.
IBM is a terrible work place if you are technically inclined, if you are a slime ball of a manager it’s perfect for you.
IBM "Midrange", aka AS/400: https://en.wikipedia.org/wiki/IBM_System_i
They were an interesting - but dead - sideline of computing evolution. I never used one, but they had a interesting object-based (instead of file based) operating system.
A friend used one of these (the shorter models) as a coffee table -- I distinctly remember how hard it was to move the big, dense metal cube in and out of his room!
My only experience on them was interacting with some API hosted in an AS/400 via FTP protocol. Felt like I was going back to the 90s.
I guess it wasn't a real country after all.
Not to mention, a ridiculously difficult problem that can't only be solved by technical breakthroughs, but also relies on legal and social advancements in order to see widespread adoption.
Also a problem that when solved completely changes their business model (not owning cars) into a different one (owning robot cars) that’s much less profitable and they have no expertise in.
As time goes on I find more and more parallels with designers and legal/policy writers.
One just builds a specification of a technical system while the other builds a specification of a societal system.
Societal systems just have much more legacy code and inertia with very little in the way of mechanisms for in house testing at scale. There's no development server, just keeping tabs on competitors (ie other countries) to see how the change in production (ie a new law introduced) had second order effects.
Ubers not been one to shy away from legal skirmishes though.
Even so, profitable self driving is binary; either there’s a human driver in the car or there isn’t. Uber is ”bold” but not bold enough to put thousands of self-driving cars on the road illegally.
That's just not true. Way before Uber IPO'ed and started aiming at profitability next year, self-driving cars was the big thing for Uber's future.
(Disclaimer, I worked at Uber 2014-18).
I'm not saying there wasn't excitement about AVs. The specific claim I'm disputing (which I've heard over and over again over the past few years) is that either company considered AVs to be the only path to profitability. The history is a bit complicated, because before the IPOs neither company was focused on profitability as a primary goal.
I agree for Lyft but Uber pretty aggressively pushed the vision as necessary for their future when they were burning so much cash on ATG.
They pushed it as necessary for their future because the future will be self-driving and anyone without access to that tech will be out-competed. That's completely different than it being necessary to their future because their present business model cannot be made profitable.
You contradicted yourself. You said anyone without access to self driving will be out-competed but that doesn’t mean it’s necessary for them. It’s extremely necessary for them. I’d much rather take a driverless Waymo than a ride with a random Uber driver that might be driving distractedly, might get lost, etc. Not to mention the robotaxi should be cheaper at scale as they can operate 24 hours a day.
"only path" is a much stronger statement than you were reacting to, which is that it was the chosen strategy.
And nothing about this deal indicates Uber is stepping away from self driving cars.
They basically structured the deal as buying stock in the company and investing $400M.
Letting them focus on their core business while still having a large financial stake in AI cars.
The big development is that self driving cars turned out to be an extremely hard problem which didn’t match company timelines (or potentially company business models, if for example AI kits can get installed in existing cars and brands for which they become the denominate call network) and they outsourced it to one of the best companies around.
"turned out"... it's been starkly obvious since day one that true self-driving is a problem that requires general intelligence to solve. All you have to do to know that is drive a car and think about what's actually required to successfully drive from A to B in the real world.
Waymo is running self-driving cars with paying passengers and no backup drivers right now in Phoenix, and last I checked we don't have general AI yet.
https://www.reuters.com/article/waymo-autonomous-phoenix/way...
Sure, you can move the goalposts and say it's not "true" self-driving until it can handle all environments and weather conditions and whatnot, but self-driving cars are here. The question is how fast they'll go mainstream.
This doesn't prove that the technology is ready or safe. It only proves that Waymo thinks it's a worthwhile gamble. This is the same place where Uber's self-driving car killed a person with no repercussions to the company.
It's obviously ready and safe enough for Waymo to roll it out (and they've been a lot more conservative than Uber here).
As for safety, all we really need is for them to be safer than the average human driver. As evinced by the ~40,000 road deaths every year in the US, that's really a pretty low bar. I'd even argue that full autonomous is/will be less risky than halfway houses like Tesla Autopilot where neither side is really in full control, but that's another story.
That’s actually a high hurdle to clear. Looking at fatalities per hundreds of thousands of miles driven humans are insanely good at driving. Using the 2018 NHTSA number of 1.13 deaths per 100 million miles driven, avg speed of 60 mph, 20 predictions per second, that equates out to an autonomous system can make a fatal mistake on average about every 120 billion predictions.
You make it sound like every mis-prediction could result in death.
"Ready" or "safe" are matters of opinion, though.
It's ready/safe when governments ("the people") decide it's such.
The technology is safe:
https://storage.googleapis.com/sdc-prod/v1/safety-report/Way...
They've been very careful about it, and they have not finished work on it, so they need to keep the same standard that they had until now.
The technology is also ready for Phoenix. The company is not yet ready for the whole world (but I have no specific insight into what's holding them back).
It's totally different from Uber.
The obvious question is, why does this work in Phoenix?
Where I live, road conditions are changing all the time. There’s new construction, new roads opening, temporary (a few hours at a B time) changes in lanes where you have to drive against the normal direction of traffic.
You can call it moving goal posts, but to deploy them generally, they either have to have general intelligence, or you need to input construction schedules, so that it won’t take certain roads during certain times. And then what? You can’t order a ride to certain locations?
Waymo has been dealing with construction since day one. If the road is entirely blocked, Google Maps/Waze is already capable of changing routes on the fly.
There will always be corner cases that automation can't deal with, but Waymo will have some human troubleshooters around for that. So do us humans, for that matter: when a car breaks down, most people call the mechanic instead of applying their own general intelligence to fixing it.
What’s the mechanism?
Eg A road is accessible from 5am to 8pm. Outside those times traffic conditions change. Humans setup cones and control the flow of traffic. 2 lanes becomes 1 lane for all traffic. And traffic is managed so that depending on what direction you’re going, your driving against the normal flow of traffic.
This is a real scenario in a moderately busy area. Times may be wrong. The road is not blocked. But you have to take direction from a person. In fact this sort of temporary traffic change is fairly common especially due to road works.
That's not moving the goalposts, those are the goalposts. Anyone can make a car that loops round a racetrack by itself, on a good day (well, not anyone, but it's relatively straightforward). The challenge is making it robust against "the real world".
Something something "self-driving-complete"...
Not human-level AI, though. Plenty of animals do fine with challenging navigational problems. Birds fly through forests, catch or evade other birds, cross continents, etc.
What am I even reading. "Nothing indicates Uber is stepping away from self driving cars". Yeah right.
Prior to this deal, Uber had a couple of billion (at the value of this deal) in a self driving car company named ATG. After this deal, they have a couple of billion in a self driving car company named Aurora.
That might not be their internal strategy, but that's pretty much the investment thesis of every Uber bull buying their stock at $100 billion market cap.
I actually poked into this a bit out of curiosity. Without doing deeper research than I have any interest in, indeed Kalanick never seems to have publicly said anything like "We are so f'ed without self-driving" (which would be a truly idiotic thing to do) and even made some references to the length of the transition. But he also made reference to things like Uber's growth being limited until Uber was cheaper than owning a car and a ton of pubs seem to have "independently" reached the opinion that this was necessary. And I also don't see a lot of fervent denials from Uber in this regard.
"If we are not tied for first, then the person who is in first, or the entity that's in first, then rolls out a ride-sharing network that is far cheaper or far higher-quality than Uber's, then Uber is no longer a thing," Kalanick said." - https://www.businessinsider.com/travis-kalanick-interview-on...
>And I also don't see a lot of fervent denials from Uber in this regard.
Why would there be? If it's helping their stock price or even neutral to it then why deny it and risk it going down?
It seems like a narrow path to try to navigate. When a reporter prefaces a question with "Next year when you've perfected your transform-manure-into-gold machine..." it seems executives either quash that nonsense, run away screaming "I can't hear you!", or make themselves liable when the TMIG machine doesn't ever work.
Does profitability equate to paying back all the VC money burned getting to this point? They were losing billions every quarter for a long time. I haven't looked at their quarterlies in a while though but Id be surprised if they were remotely right-side up.
No, because the VC money is equity rather than debt. Profitability doesn't include "make sure the owners get some cash." It just means you made more money than you spent this quarter.
Investment rounds aren't loans, they're... Investments. You're selling them a stake in the company, not borrowing money. I'm sure there are exceptions but that's generally how it works.
No, it means a profitable quarter.
It may not be the strategy, but it is definitely the only reasonable path to profitability where the drivers will make at least a reasonable income and the cost of a ride is competitive with a cab.
Either ways, next year is pretty close. Perhaps I'd be proven wrong.
Only because they have burned so much VC. Both Uber and Bolt charge like normal taxi companies in Stockholm and can therefore afford to pay their drivers at the same time as making a profit. And as far as I know Bolt is basically already profitable, they just copied Uber's business but reduced the administration costs and did not blow a ton of cash on self driving cars. Having their HQ in Tallinn rather than SF also helps.
There is definitely a market for a modern taxi company, but it won't be as cheap as Uber is in the US. And Uber has too large overhead.
Hard to pay your execs and engineers insanely bloated Tech salaries on Taxi profit margins. It's a bubble ready to pop.The guy saying they will profitable in the next year is delusional.
"The guy" - the CEOs of both companies? You're essentially stating that the CEOs of Uber any Lyft are lying to their investors, which is a crime. That's a serious accusation and some actual evidence is needed.
No, he's stating the CEO is delusional. It's not a crime if you honestly believe what you're saying.
(Oversimplification, of course. I'm fully aware of corporate obligations around market disclosure, profit forecasts, etc).
This is hilarious, honestly. Who was arrested at Helios Matheson with the joke that was MoviePass?
What strategies do rideshare companies plan to employ to reach profitability: increasing prices, cutting costs, etc? And what will the implications be for riders and employees alike.
My assumption would be both. Taxi companies have way lower cost structures. They certainly don't employ lots of software engineers. And there's no reason to think that Uber should be significantly cheaper than a taxi ride. Scale buys you something but probably not a lot. (And scale has some costs too.)
You underestimate what you can extract from making the market larger and more efficient, which taxis could never do. Why don't people barter instead of paying huge sums to Wall Street to intermediate?
Bartering in this problem space is hard. Let me call up 5 of my closest friends in my city, to see if they want to leave their home, drive me from Point A to Point B, then go home, right now.
In part, because, on an individual basis, Wall Street's skim is tiny. It's only big in aggregate.
"If we are not tied for first, then the person who is in first, or the entity that's in first, then rolls out a ride-sharing network that is far cheaper or far higher-quality than Uber's, then Uber is no longer a thing," Kalanick said." - https://www.businessinsider.com/travis-kalanick-interview-on...
Profitability maybe, but the obscene valuation was based on Uber having a self driving fleet one day, at least to my understanding. Many businesses are profitable and even much more profitable than what Uber might be in a few years time and yet they have never come close to any valuation such as Uber has seen.
Why are you repeating this falsehood? What facts do you have to present that Ubers current valuation is majorly driven by a selfdriving future?
What is the strategy to get to profitability then? Have only heard of the self driving route before.
Can you point to some sources to back that up? I have always been under the impression that self driving was the end goal that they were all working towards.
Everytime someone mentioned that idea I thought it made no sense. The Uber strategy is to delegate all the cost to a 3rd party (the driver and their own car). The idea of spending billions of dollar on hundreds of thousands of cars around the world and maintain them hardly seemed like what they were setup to do. They wanted the exact opposite. Externalize all cost, labor, material, etc. and just develop a software that generates them a % of profit. This would be an entire 180° in terms of business strategy.
So if they invent the first reasonable self-driving car, they can sell those and provide service for them. Ideally TCO would be lower than a human driven car so basically a third party could buy and manage the fleet for Uber and Lyft. Neither company would need to own the cars to reap the benefits of making them.
They could, but that’s a completely different business to building middleware to facilitate unlicensed taxis. It’s like saying “AirB&B will help people rent their spare rooms out until they launch their own hotel chain.”
How do drivers make any money while keeping the overall price less than a taxi? Factor in fuel and maintenance, and drivers are already getting pretty screwed on these platforms.
They don't. Which is why Ubers cost only slightly less than normal taxis in markets like Sweden where Uber has to follow the same regulations as their competitors. The low prices they have in markets like the US are not sustainable.
Prop 22 voters think otherwise... but yeah they're delusional.
Absent self-driving--which was always a pipe dream--the overall price will not be materially lower than a taxi. Perhaps people will prefer the service for other reasons. Perhaps they'll go back to transporting themselves however they were pre-Uber.
Taxi medallions used to cost hundreds of thousands (SF) to more than half a million (NYC). Nowadays they still cost tens of thousands in SF and hundreds of thousands in some areas like NYC.
Also, more efficient use of time from app-based routing.
I'm not convinced. Do you have evidence that "more efficient use of time from app-based routing" actually results in drivers making more money?
"app-based routing" is a mirage anyway. When you request an Uber ride, your request is advertised to drivers in the area, who can choose to accept it. There is no routing happening.
Sure, the driver then drives a route given to them to by the app, but so does everybody else, just in that case, the app is Google Maps.
AFAIK, the only routing cleverness is that the end of a shift, drivers can request to only take passengers who are heading in the direction of their own home. That's useful.
Medallions alone should be enough to make a difference. The efficient use of time probably has much less of an effect on earnings, if true.
I'm also not terribly convinced of this. Medallions are a one-time fixed cost. This should have little to nothing to do with long term prices, especially given that the prices of medallions have gone down significantly.
That seems disingenuous and intended to save face. I liken this to the Soviets claiming they never lost the space race after the US moon landing because "landing humans was never the goal" and they denied ever building a lunar lander. I think the kids these days call it gaslighting. Statements from key executives at uber like Meyhofer who said “The ultimate north star for the company is Level 4 autonomy” are proof of this intent.
Yeah the more I look the more I find proof that self-driving cars has been seen as an existential threat by Uber/Lyft for years and years with Travis Kalanick (ex-CEO) saying it repeatedly. Here's an interview from him on exactly this topic. https://www.businessinsider.com/travis-kalanick-interview-on...
If there are competitors (e.g., Waymo) that have self-driving that's significantly cheaper, then Uber and Lyft can't compete. So it's a huge potential threat to their businesses. And if self-driving makes rides much cheaper, that can significantly boost the size of the market. But that does not imply that they can't be successful and profitable businesses in a world without self-driving cars, which is the frequently-made claim in the top post.
I agree that saying self-driving is a threat is very different from saying that your business will fail unless you create self-driving cars.
Also, saying that Uber/Lyft need to create their own self-driving cars is ignoring the fact that other companies that create self-driving cars are, and will likely continue to offer rides in their cars through existing ride sharing companies.
> other companies that create self-driving cars are, and will likely continue to offer rides in their cars through existing ride sharing companies
That remains to be seen. If a company is able to create self driving cars, what's the value add provided by Uber and Lyft? Having a middleman TNC platform skim some profits doesn't really make sense from that perspective.
Compared to achieving reliable and cost effective autonomy at scale, building a ride sharing app is trivial, and customers have indicated that they're willing to install another app if the fares are cheaper. The complexity of getting a self driving service is high enough that there probably will be only a few entrants in the market, unlike the current value add provided by the TNC apps that match customers with a large pool of individual providers.
Uber seems to be expanding into food delivery and other logistics, where trust is still valuable so a middleman can be useful, but it’s hard to say how well it’ll work.
Realistically though it's about as much of an actual threat as teleporters are.
Can't speak for Lyft leadership, but I think Travis and Dara have very different perspectives on self driving.
As for the existential threat narrative, consider that there are many places that sell cheaper/better food than McDonalds, and yet, McD is everywhere. I don't think technical merit alone is going to win the rideshare pie.
Cheaper than a $0.39 hamburger? Where?
What currency are you using? In USD, the cheapest item on the menu that I can find (apple slices) goes for $0.57. A big mac combo goes for around $10, while a subway footlong goes for about $7.
$10 is also a pretty wrong number for "hamburger". Looking at this ordering form, I see $2.49 for a mcdouble and only $1.39 for a basic hamburger. And the mcchicken is $1.19
(Though wow, mcdouble was half that 2-3 years ago. Big mac hasn't really changed.)
I do recall an article from a few years ago about how the basic McD burger was arguably the most bang-for-your-buck in terms of calories per cent, and I agree it's hard to beat (though not impossible, e.g. the costco hotdog combo costs a measly $1.50 and packs ~3-4 times more calories)
But yeah, a lot of the McD classics and new offerings that one might purchase for a "normal" meal are actually quite expensive nowadays. On a side note (pun intended), I suspect that there are some pricing structure shenanigans going on, because fries are clearly marked up way more than burgers.
Well, if they were only afraid of them, rather than needing them, they're in decent shape, 'cause the last few years have kind of given the impression that self-driving cars aren't coming any time soon. They've been "90% here" for at least ten years but that's not enough.
Or maybe they're just playing both sides. That way they always come out on top.
Within Uber, it was categorized as a "big bet", as opposed to a core vertical, like ride share/eats. Having now shed new mobility, elevate and now SDV, It looks like Dara is really set on focusing on the core verticals now.
Big bets and promises for the future was all they had. Trying to be a solid profitable business in a dying industry isn't going to work out well for them.
Wait, how is ride-sharing / cabs a dying industry? People are always going to need to get from point A to point B, and if anything, the younger generations are less likely to own cars. Public infrastructure is never going to fill the gap except in certain urban areas.
>the younger generations are less likely to own cars
There's nothing inherent about this. If the promised alternatives to owning cars turn out to be BS, the younger generations will have to own cars whether they want to or not.
A much more durable change is the shift in social norms towards intoxicated driving being unacceptable under any circumstances. Just 40 years ago the DUI threshold was .15 and even then the penalty was light. Today it's .08 with talk of lowering it, penalties have been ramped up multiple times, and nobody sees DUI enforcement as a technicality.
That's a pretty big opportunity for any firm that gets you home from the club at 3 AM.
I don't disagree with that but is the shift in drunk driving norms really enough to shift from owning cars to taking taxis? In my experience, most people don't primarily use cars to go out drinking--including among those in a position to just take Ubers instead.
Personally, I think drunk driving isn't that much of a factor: drinking is a relatively short term kind of planned activity. What I think has a much bigger impact is long term stuff, like the desire to live downtown and/or housing prices (with downtown condos and midtown townhouses being more affordable than big detached homes in burbs)
I know families that are rather against the idea of owning a car and arrange housing around that consideration. They prefer to rely more heavily on public transit and the occasional car rental or cab/rideshare.
I wouldn't say that, no. Other factors might reduce car ownership, of course, but I meant only to address the larger question of Uber's potential market. Another major use case is the airport, because airport parking has become egregiously expensive and cost-cutting favors rideshare infrastructure over parking (which is a good thing land-use-wise). There's also "transit in taxi out" trips because it's cold at night and people are tired and don't want to walk/wait/transfer, which grow as public transit is expanded, another trend that is likely to continue because the reduction in air pollution from electrification makes urban areas more desirable, as well as various secular trends pointing towards urbanization that may or may not continue (some recently reversed by the covid-19 pandemic) -- environmentalism in particular seems like it won't go away.
In any case, plenty of people are enthusiastic about a world where owning cars is less popular, so it's a common refrain.
Sometimes I wonder if people talking about this "inevitable unprofitability" have used ridesharing recently. At least around SF, prices were steadily going up for the past couple of years.
The younger generation is always more likely to go to cities. As they get older they move out. And buy cars.
Are they willing to pay the actual cost, though?
I know that where I live I've started blanching at how much it actually costs to take an Uber, which is now approaching what taxis cost.
cost is fine if you don't take one every day, which in a remote work world is not at all a stretch
but that would effectively limit Uber to a smaller, pre-Uber sized market. Which would probably be news to investors.
> the younger generations are less likely to own cars. Public infrastructure is never going to fill the gap except in certain urban areas.
And Uber/Lyft work reliably only in urban areas. "Younger generations" will inevitably grow older and may want to move out of tiny, expensive city apartments to larger accomodations outside the city. Guess what. Public infrastructure fills the gap much better than Uber/LyfBolt do (at least in Europe).
Elevate is also gone ? Didnt read about that
Kalanick said that self-driving was “existential” to the company.
But in context, he didn’t mean Uber needed self-driving to exist. He meant Uber can’t be left out if self-driving becomes a reality.
In that sense, the sale seems pessimistic. They no longer see self-driving on the horizon, or they aren’t on track to deliver the technology themselves.
They’ll have a major share in Aurora after the fact. They just are focusing on their core competencies, I assume. To compete in the self driving space, you need to have 1) a crap ton of money and 2) the patience to wait 10 years before you see any ROI. Uber is not structured to do that this late in the game. They’ll probably get a preferred partnership for licensing the tech once it rolls out, if Uber still exists in 10 years.
Do they get ROI on rides or Eats? I was under the impression they were still pretty substantially subsidizing prices on those, though maybe that's just out of date information.
It's not about subsidies as a business model so much as a war of capital attrition with weaker competitors, which is necessarily not endless.
Current competitors may go out of business. If there really is a niche, however, seeing off current competitors won't be the end of it. There's always someone else, who may not have burned so much VC.
The other option is that there will never be a sustainable business here.
If so then why would VC's fund new competitors? The answer is they would not. VC's may throw their free money at new unquantifiable opportunities, but for this niche the financials will have become well known. In other words there would be less risk premium.
VCs are looking for firms who can build "moats", i.e. commercial positions within markets that have built-in advantages against potential competitors. Sometimes this is some technological feat that is difficult to match, like Uber have been promising. Other times this is unethical business practices, like paying employees less than minimum wage. Such moats may not last very long, but the concept appeals to VCs.
The phrase "war of capital attrition", however, will send capable VCs running for the exits. If they liked that sort of thing they would have been bankers.
That claim was infinitely repeated on internet message boards but never acknowledged by the company or taken seriously in the industry or by investors. That entire division was an expensive moonshot. Uber's core strategy and valuation have always been rooted in being a taxi company with human drivers.
So whats the big plan to lower overhead on rides? I looked up their last quarter and it was billions lost yet again. This doesn't seem sustainable, just shoveling money into the furnace quarter after quarter like that.
The big plan is a war of attrition to capture a monopoly/duopoly status for taxi services within a given country. Acquire all competing taxi companies or price them out of business.Once that is done, raise prices. If competition arises, lower prices again until they go away. In some senses, its just like Amazon.
How long can they afford to fight it? It's not hard to spool up a rideshare competitor per say. When they pulled out of Austin several other rideshare companies popped up in their wake. If prices got exorbitantly high for the average user, I wouldn't be surprised if a company came in undercutting uber.
Agreed - its "an" investment strategy, not necessarily the best one. And if you have Softbank's backing - maybe a valid one? Kinda works/worked for Walmart/Amazon.
Selling or shutting underperforming divisions (like ATG, air taxi), reduced competition in the space as players consolidate, higher prices, cutting expansion/marketing costs.
They can sustain it as long as investors are willing to shovel money into Uber... and no longer.
Uber's true mission is a charitable effort by SV capitalists to subsidize the cost of taxi rides. Of course they're not meant to make money. How silly to think so!
It’s actually charity by SoftBank and Saudi Arabia at this point, but yes, this is what it actually seems to be.
Amazon is also an investor charity to give us all one day shipping since the investors show no sign of wanting profit margins above 1%.
Uber don't need to invent self-driving themselves so long as whoever does invent self-driving cars sells them on the free market. Uber has a strong brand and loyal customers, and they could make out like bandits even if they have to buy the tech from someone else.
Their real problem is if some other company perfects self-driving cars and doesn't sell them. Perhaps Uber have decided that's unlikely, either because no one will or because a company that isn't looking at ride-sharing will.
Uber has a strong network effect, but I don't think there's actually much brand loyalty.
I'm not sure how strong a network effect other than in a local area. I'd be pretty sure that (in normal times) their primary market isn't business travelers flitting from SF to NYC to London but residents of $CITY using Uber Pool for commuting or Uber to go out with friends on a night out to a bar.
Contrary to that, local profitable markets like Manhattan have had significant entrants eat away at Uber’s market share with companies like Juno scooping up the best customers and drivers. The same thing happened in many cities in the early 20th century before streetcar operators managed to get jitney cabs (the unlicensed pool taxis of their day) banned because too many teens would borrow the family car after school and drive in front of streetcars on profitable lines and scoop riders for nickels instead of dimes.
Uber and Lyft have a de facto duopoly amongst those hopping off planes because it’s hard to know that there are any other local options available.
And for that matter, I often take a cab from the airport because it's not my money and I can't be bothered to mess around. Uber and Lyft are often more trouble than they're worth.
But if there's a local alternative known to be better for going out with their friends, you can be sure that a person used to going out regularly with their friends will know what it is.
Perhaps negative brand loyalty, to be honest.
It seems really premature to make any declarations about brand loyalty when Uber is still majorly subsidizing consumer prices with a VC cash bonfire.
The big scare at the time Uber started it's own unit was that "once Waymo or another competitor has self driving cars, it will be easy to offer a ride share network and compete with Uber"
that sounds unlikely. self-driving isn't like a social network, the ones end up building it would've invested tens of billions - so basically Amazon or Google. I find it hard to believe they would want to sell that technology to Uber. If anything, Google has large investment in Lyft, more likely they will acquire Lyft to deploy their own driving fleet all integrated under G Suite.
You don't need to own the tech to use it, anymore than they would need to be manufacturing the cars they're using.
OR: they finally accepted the fact that self-driving cars aren't realistically happening anytime soon, at least to the level needed to entirely eliminate the driver cost.
That, and they probably realize they have enough market share at this point to be able to (long-term) raise prices instead of lower expenses.
Even if that were the case, selling your self-driving car unit might still make sense.
If you assume that self-driving cars eventually become a commodity, Uber and Lyft could just buy them rather than develop the tech in-house. The key strategic bit of the landscape to own would be the app that everyone uses to book them.
No matter who makes the driverless cars, they will still need a network to run those cars on. I think Uber/Lyft area better off allowing thirdparty self-driving cars on their network. Waymo on other hand is trying to build its own network, but that is going to be extremely expensive proposition.
IMHO being the only one with self-driving car technology would be a much better moat than having 'a network'
London has Uber, Bolt, Kapten, Ola, Viavan, Hailo, Wheely, Xoox and that's just a single city - so the ride-hailing-app tech isn't a moat. A self-driving-car business doesn't need any drivers on board, so driver recruitment isn't a moat. And Google already have Google Maps on a big fraction of phones, so app installation is scarcely a moat.
Of course Google will have plenty of opportunities to shoot themselves in the feet, so it's certainly not a done deal.
I always considered that a dubious business model. It's outside of Uber's core competence, and it won't lead to superprofits/a moat for very long, since they'll have competitors. Sure, it lowers their costs, but once a second (third, etc) competitor has SDCs, Uber will be able to charge a lot less as well, because competitors will have lower costs too. It's not the golden ticket to profitability.
I've been long hoping, for their own sake, that they'd sell off the SDC division, since it was hurting their balance sheet anyway. Glad they've seen the light.
It doesn't have to be developed by them. Just like the don't make their own cars. If some car company makes a reliable level-5 car Uber can just offer a deal with owners of those cars.
Was there anyone who actually still believed that story? I mean, it should have been pretty obvious for at least a few years but did someone still believe it? (Less snarkily, they do still have a financial interest but yeah.)
Maybe they've decided that Uber's future lies in letting someone else solve that hard problem of self-driving cars and Uber figuring out how to line up the capital to lease self-driving cars from that winner.
Clearly not since they're still sitting at a $95B market cap after this news. Or perhaps just nobody else has realized this, so why don't you short the stock?
When you look at the numbers, you actually see how Uber actually paid Aurora to take ATG.
The headline is a little decieving. Though Uber sold their self-driving division to Aurora, Uber is taking a stake in the company so it seems like a partnership.
This self-driving system promise keeps getting delayed! It looks to me this is a case of a donkey led by a carrot on a stick (investors are the donkey here). What I never understand is what happens when a new situation that the system did not previously train for happen? what if this situation is deadly? Will all the Uber cars decide to jump off a cliff because of a wrong signalization or bad weather? What will the Teslas decide? Nature is very capable of creating 1 thing and multiply it, but it does not. The reason is it is more secure to have variation. If a problem happens, it will not affect every member (same what is happening with COVID by the way) Creating a mono-brain will soon prove disastrous. You just need a previously not seen situation or a crazy guy that gets imaginative and that will induce the death of hundreds
> What I never understand is what happens when a new situation that the system did not previously train for happen?
This happens to human drivers all the time, and humans are really bad drivers even in situations that have happened before.
I fully expect AI drivers to be safer than human drivers.
> I fully expect AI drivers to be safer than human drivers.
Don't you know the rules of AI claims? You always add "in 50 years" to every statement.
The biggest problem is accountability: who pays when someone dies. Once that is worked out, it'll have a shot.
Eventually sure - but in the start one road block I see is not just total numbers of accidents but also the way it fails. Human failure modes are relatively well known and we build our infrastructure around them - AI not so much.
One pedestrian death due to cars occurs on average every ~200 million car-driven miles. Is that really bad?
(the math is based on # of US miles driven in 2017 divided by the number of car-caused pedestrian fatalities in the same year)
I mean only counting the 5000 pedestrian deaths ignores the 137000 emergency room visits by pedestrians due to cars. Many of which probably involve massive suffering and lifelong disability. It also ignores the 38000 car occupants that die every year due to accidents. And the 3,000,000 injuries to car occupants each year of which 2,000,000 include permanent injuries.
So cherrypicking the smallest number seems very disingenuous to me.
Is that right? That seems phenomenally bad. I feel like I have normal American driving habits and I drive a standard 10k miles per year...
So I'm going to kill a pedestrian every 100 years? Assuming I drive normally for ~50 years, 500k miles, I have a 50% chance to kill a pedestrian?
Did they change the number? The comment says one pedestrian death per 200 million miles right now, which is in the right ballpark and gives you a chance of 0.25%
yes I updated the number. Sorry about that.
1 out of 6 traffic fatalities is a pedestrian. Roads are designed poorly, ice exists, cities are built on steep hills, and people walk at night on the side of county highways.
only assuming a random naive model. how careful you are and how dangerous the location also matter.
Yeah, but it's also only counting pedestrian deaths.
25%
It's the loss of agency to machines, not the raw numbers that scares people. Why do you think so many people are unduly spooked after airline accidents due to mechanical or computer error?
It’s the fundamentally different kind of danger from machines. They make mistakes we can’t understand.
I can understand a person being distracted, reacting slowly, not seeing me, etc. I can also mitigate this risk. Crossing the street I can make eye contact with a person and be really confident that they aren’t going to run me over.
Walking in front of a robot? I have no internal model for the kind of mistakes they make. It seems equally likely in any circumstance that they’ll make the mistake of running me down, and there is nothing i can do to avoid it.
You do develop an internal model for the behaviour of machines, though.
Examples:
Elevators where the doors take extremely long to close on their own.
Elevators where the doors close too soon after opening.
Elevators where the limb sensors are not very responsive.
Cars that take a couple of tries to start in cold weather.
Automatic doors that are a little bit slow to open.
Etc.
Powered doors are one dimensional machines with perhaps one binary sensor and a motor driver, a person could implement a state of the set automatic door in an afternoon.
You would also have to be rather clever in order to get an automatic door to injure you.
A car is also a rather easily understood dynamic system, but that isn’t an issue either. We are all quite frequently in the position of being at between a few seconds and one heartbeat away from a car killing us... walking down a sidewalk, crossing the street, driving on the highway...
We understand pretty well the dynamics of a human driver. Even the drivers that make terrible mistakes we understand something of hope and why.
A robot driver though... the terrible mistakes it makes are mysterious and terrifying. They make the kinds of mistakes a human never would (and few mistakes a human would make).
but those dont happen at 60+ miles per hour
Okay, let's use another example: lifts don't move until the doors are closed. If the lift thinks the doors are open, it won't move. If it did, it would be an out of control situation just as quickly as a car misbehaving at 60 mph is. The expectation is that autonomous cars must be cautious.
my point is when we develop expectations and mental models of how machines work, we can use those to adapt and feel safe and in control. but that's less possible when the machine is typically moving at 60 mph and the mistake can happen without warning
Like elevators where the brakes fail. If it's infrequent enough, we tend not to think about it even if it does still happen from time to time.
I see your point but for me at least, my gut feeling is the autonomous cars are just scarier because they're moving at 60 miles per hour. Maybe it's irrational but this whole discussion isn't about rationality its about gut feeling
Yes! You can control for human risks in the tail scenarios. You can't control for the tail scenarios in AI because you don't know what on earth they are, they are a nasty surprise, thrown up by a model which, in the tail, we as engineers don't fully understand
I would say that, when I think about it occasionally, I am kind of scared by the possibility that someone will hack into all the self-driving machines (from a particular vendor) and make them all do something terrible at once—before anyone has time to deploy countermeasures.
It went as an unspoken, unquestionable assumption that telephony was the right model for data networking. - Van Jacobson
Maybe a single 'driver' agent in a free-moving ground vehicle in close proximity to vulnerable actors and objects is not the right model for scalable, efficient, safe personal transport?
Which is ironic, because this is the kind of automation they've been building into planes for years and consequently air travel is safer than it's ever been.
737 Max fiasco notwithstanding - it's worth mentioning that MCAS had nothing to do with flying the plane, only making it feel more like the 737NG to its human pilots
> it's worth mentioning that MCAS had nothing to do with flying the plane, only making it feel more like the 737NG to its human pilots
Not entirely true, the original issue is that the engines are bigger and the center of gravity is moved forward ( because the 737's are too low), thus the planes tend to stall. MCAS is there to correct the stall so that they feel more like 737NGs, but the issue is forcing bigger engines on an obsolete design.
How much machine learning and black box decision making does autopilot in a plane do?
My understanding is none, or nearly none. VNAV follows GPS waypoints with well described, manually designed behaviors for things like climb rate, bank angle, etc.
Autoland isn't machine learning either, and has such a large list of requirements to be allowed that it's basically like designing a self driving computer for a train.
There’s one death per hundred million miles driven. One per million would mean one in 2 drivers kill someone in their lifetime.
corrected- thanks. I hate being off by two orders of magnitude.
Motorists are IIRC the third-biggest killer of young people.
Depending on what you consider young, it's the #1 cause for ages 15-24, #2 for ages 5-14 in the US.
That is true, but what if it was the same car (software) killing every pedestrian? Right now every situation is unique, but being judged based on who wrote the software is the worst case scenario for self-driving regulation IMO.
Pedestrian deaths are not the only bad thing that can happen when driving.
Human drivers can adapt to a new unseen before situation.
And software propagate learned knowledge perpetually, whereas each human has to learn it over and over again.
On the other hand, software can make mistakes at scale whereas each human makes mistakes differently.
It’s not so cut and dry.
While potentially true, it's a trade off it seems proponents are willing to take.
The same kind of trade off happens with other modes of transport with very rigid constraints on movement, things like trains. A train can keep going, stop, or maybe with enough time reverse if there's tons of time and a strong need.
We don't stand around worrying saying "BUT WHAT IF SOMETHING'S ON THE TRACKS?!?", because the answer is "well then, the train will hit it. That's the tradeoff for being able to move freight so efficiently."
Sometimes they try to drive through floods. They also seem to randomly run red lights. Also, what's with the way they keep veering out of their lanes?
I think what this boils down to is that the volume of vehicle-related injury/death will decrease, and the few remaining cases will be really strange and unlikely to have occurred with human drivers.
Some human drivers can adapt to a new unseen before situation. Many can't in my experience.
I also expect AI drivers to be safer than human drivers ... in about 50 - 250 years.
well the one issue that should be obvious to all is that self driving cars suffer worse from the problem people suffer from, signage and markings on roads is still either incomplete or in some cases in error.
People screw up every day driving and the lives lost is very high but going by how my Tesla drives I am going to say impede traffic by slowing down. My car does slow down, safely, when it gets confused or contradictory feedback, and I am curious what the in beta FSD software will do. Their new technique looks vastly improved.
Traffic aware cruise control which many cars have along with limited lane keep assist can go a long way to making the drive safer but many don't use TACC except on highways
Look at it this way, its coming. However it is all one sided because again I emphasize, we need better and more consistent signage and road marking.
US wise, the HOV and Express lanes are probably the most perfect use case for self driving cars. Get into the lane and it should be hands off until exit.
> What I never understand is what happens when a new situation that the system did not previously train for happen?
I think most people ask this kind of question because they assume the entire system is some kind of big if-else-if-else statement. The question about swerving and hitting the kid or the grandma follows the same logic.
What we have to realize is that a neural net can and will react to things it hasn't specifically been told about using heuristics and "similar" stuff, just like human drivers. (see video of Tesla FSD avoiding a deer recently [1])
This is exactly like a new, or student driver. They have not yet encountered every possible situation, but they're able to apply reason and logic to new things and the more they drive, the better they get. So although they've never seen a deer jump out, they know it's good to avoid things when possible BUT NOT to swerve into oncoming traffic if there is traffic. So even a new-ish driver who has never seen a deer on the road will do a decent job of slamming on the brakes and swerving if save to do so. Same goes if a tire bounces towards you on the freeway, etc. etc.
Tesla are working towards billions and billions of miles of this kind of "learning driving", where the system keeps getting better and better.
And even after those billions of miles there will be situations it has never seen before, but the required response will be extraordinarily similar to things it has seen before. Chances are it will do a better job than you or I, because it's reaction time is close enough to 0 not to matter, and emotions won't cloud it's judgment if something really extraordinary happens (plane lands on highway, flood washes away road, etc)
This is exactly like a new, or student driver.
While I agree with your overall point, this is not true. AIs often fail in very inhuman ways. Their "thought process" and learning capabilities resemble ours only superficially.
That said, humans also make a ton of lethal mistakes on the road and I do think we're getting close to a point where it might even out on balance. But people are more forgiving of relatable human failure modes than bizarre machine ones.
That video isn't an example of this. Tesla's statements about fleet learning are pretty much theoretical - they're talking about a system they want to build someday. Right now, their motion planning and prediction is all hardcoded robotics.
See Karpathy's talk here: https://medium.com/self-driving-cars/annotated-karpathys-aut...
Uber's has. But Waymo is currently giving self driving rides. By some definitions self-driving is here.
Obviously there is still a ton of work to go from easy-to-drive areas to global but it seems like that is a huge notable step.
I am convinced that it takes another 100 to 200 years before we really have smart self driving cars. Maybe it will never happen as I think it will be much more likely that we have self flying drones before long before. Airspace being regulated heavily and thus much more predictable for a machine.
100 to 200 years is a very long time. Look at the exponential nature of technology, and even deep learning. Only in 2012 had the image network reached 90% accuracy levels, and now we have all sort of networks for all sorts of things, including style transfer for videos.
There's nothing to suggest technological growth will be exponential forever. It's just as likely we've gotten all the easy stuff the last century and now everything else is much harder.
That supposes that we've reached the flattening upper section of the S-curve of technology, right about now. There's nothing that suggests that. Various techs do reach that stage periodically, but nothing indicates that technology in general does, quite the contrary.
And what the past sees as hard often becomes less hard (I'm not going to say easy, because technology is never easy) in the future thanks to the accumulation of knowledge and increased speed of processing data.
The s-curve of technology is not meant to be applied to universal technological improvements in a civilization. It just doesn't work.
> And what the past sees as hard often becomes less hard (I'm not going to say easy, because technology is never easy) in the future thanks to the accumulation of knowledge and increased speed of processing data.
It doesn't matter how fast you process data if you can't physically alter the scenario. No amount of data can make the carnot cycle more efficient or reduce entropy.
I'm sure the Romans thought Europe would never enter the dark ages too.
Maybe so, but there isn't much evidence of a slowdown while there is a lot of evidence of acceleration, such as time to new tech, re: hunter gatherer to agriculture to industrial revolution to information revolution to potentially genetic/biotech/AI revolution in the future.
That jump happened because of groundbreaking discoveries in physics, chemistry, and biology. We are reaching the ends of that, though.
Particle physics has been at a standstill for decades. Fusion power is still a pipe dream. Global warming continues to go on without a technological answer in sight.
False AI promises are not evidence of anything. Nothing AI is doing now is genuinely artificial intelligence. Biotech? We still can't cure cancer.
We are getting more and more evolutions and less and less breakthroughs. That, to me, is slowing down. That doesn't mean things won't get better - just not as quickly, or as thoroughly.
Immunotherapy for cancer is actually making great progress.
https://news.harvard.edu/gazette/story/2020/02/u-s-life-expe...
Most other progress depends on cheap energy and governmental policies, which we’re getting again.
https://noahpinion.substack.com/p/techno-optimism-for-the-20...
Like your article says, most of those deaths are things like less people smoking, and small increases in treating/early detection of specific cancers. But the underlying problem hasn't been solved. That is an evolution, not a revolution. Which is my point - less and less revolutions, more and more money spent, for small evolutions.
> Most other progress depends on cheap energy and governmental policies, which we’re getting again.
Which, again, isn't scientific progress. If it's a sociological problem it's not really a scientific breakthrough - it could have been solved decades ago if the will was there.
In fact I'd say we can gain MORE from sociological improvements, not pie in the sky science.
The point is that it's unknown what will happen in the future. In 1889, the US patent office commissioner had said that we should close it, as "everything that can be invented has been invented."
Of course no one can predict the future, that's true of anything. Not really an argument. We can only go by previous data - and previous data shows that we are anomalous, a tiny blip throughout time.
Yes, like the other day I pointed Google Lens at a QR code! The AI dutifully recognized the QR code … as a QR code, and promptly searched the web for other pictures of "QR code".
The big problem is that the populace do not understand statistics. As soon as a self driving car kills someone the witch hunt begins and the whole fleet needs to be retracted until investigation is over. Can your image network solve this problem?
It's like nuclear power. It just has to continue regardless of how many deaths since the number of deaths will be lower than other methods, in this case coal and fossil fuel deaths, and in the case of self driving, manual driving accidents.
and like nuclear power, it is the politics that will kill* it, not the technological progress.
* - Yes, Nuclear is not dead, but is not being used anywhere close to the potential it has i.e. effectively dead.
There is a push to upgrade road infrastructure with sensors so might be sooner. You're totally right about regulation though - engineers are only solving a "technical" problem, but what about the non-technical ones? Where is the brigade of lobbyists needed to get something like this through legal?
In 100 years the self-driving cars will be on Mars.
I think humans are really poorly prepared to make accurate "100" year predictions.
Otoh, we're pretty bad at 10 year ones too - the original self driving hype being a prime example. Lots of otherwise intelligent people fell into magical thinking there.
Car companies doesn't make their own airbags, they buy them from Autoliv. They don't make their own brakes, they buy them from Brembo. They buy transmissions from ZF, tires from Michelin, ECUs from Bosch, etc. Why should car manufacturers develop their own self driving technology instead of letting others, like Bosch [1] or Veoneer [2], do that and buy it from them? Spread the investment over more cars, it makes sense.
Car companies do that for the really standard, commoditized stuff. But they try to do it on their own for the cutting edge stuff. Toyota developed their own hybrid technology, and it paid off for them big time. Having a better hybrid system than anyone else gave the Prius a serious leg up in the competition. Honda's done similar work with variable valve timing, GM did it with automatic transmissions, etc.
I think, though, that that's a tangential point, because Uber is not a car company. As others have theorized, Uber has an interest in developing it themselves because self-driving represents an existential threat. They have absolutely no defense against someone else using self-driving to do to them what they used gig economy labor pricing to do to taxi companies.
To answer your question: presumably it's easier for a self-driving car company to offer "ride sharing" than a ride sharing company to offer self-driving cars. The power imbalance could be an existential threat to Uber or Lyft if it were realized.
Exactly, if a OEM or partner figured out how to solve self-driving; they would go to market first. I know people like to say its hard to build a market for drivers and passengers like Uber and Lyft; but if you solved self driving, then you should have a good price advantage to build a new market like Waymo is doing
Plus the market would no longer be two-sided, which significantly reduces Uber, et al.’s moat.
Having used Bosch software I sincerely doubt that organisation has the capability to innovate in anything software related.
because they have something tangible on their books producing value: the line for the cars.
without the self driving vertical promises, uber is a middleman with a booking app, and they're one squeeze away from tesla or waymo into irrelevance.
Another angle for car companies to develop their own tech might be (claimed) competitive edge.
Not many car manufacturer would market the airbags, breaks or tires as superior. However, car makers that develop their own engine constantly brag about how efficient or powerful their cars are compared to their competitors (e.g. Mazda and their Skyactiv, Toyota and their Hybrid).
That model really only makes sense if there are multiple compelling autonomous driving products.
If a company builds an autonomous driving capability and is able to hold on to a monopoly over it, they would either simply drive uber out of business, or license the tech to them in a way that would leave them with very thin margins.
Is it possible that just having an app doesn't make you a tech company? Anyone else ready to think Uber might just be an overvalued taxi company?
I read somewhere that there are different levels of "tech" company. A truly tech company would work without any human intervention for multiple days without breakdown. A non-tech company involves human in every tasks required for work. Most of the companies fall somewhere in between.
Ride services/Food delivery company cannot function without humans.
Are there any companies that work for days without human intervention?
Online digital content, like paid courses or an author selling their own books, may come close once a sufficient amount of marketing material is out there. Email funnels can be automated, blog posts can be scheduled (and recycled), payment gateways and whatnot can handle the actual transactions.
Simple apps and games don't necessarily need to keep getting better all the time; a successful one could likely coast for some time without needing much intervention.
google search, netflix, facebook, gmail, roomba, apple, microsoft, slack, youtube, AWS, wordpress, github, jira, the list goes on
I'm not convinced any of those would last even hours without an army of SRE's holding them together.
They all have support. True. However at my gig at my previous Job I have gone many opslead shifts without incidents. In fact most incidents were because someone added something new. During code freeze, we coasted multiple weeks without incidents.
SREs job is exactly this. Prevent future incidents. Prevention is much cheaper than cures (mostly)
Blogs for sure
So I guess SpaceX, Tesla, Intel, Apple aren't tech companies.
Agree with this rhetorical question. What mlboss said should just be "automated company" not necessary a tech company.
With that definition, it's also unreasonable to me that 100 years ago, there wasn't any tech company
the definition of 'tech' has been hollowed out for a while. Almost all software 'tech' companies are actually just market-makers or middlemen rather than actually driving deep process innovation that delivers extreme efficiency gains. Uber, Airbnb's etc.. model is basically regulatory arbitrage by externalising costs that taxi companies or hotels have to take on themselves to the detriment of workers and the environment.
You’re assuming Uber behaves worse than a taxi company, but that’s giving too much credit to the taxi company. They’re just a mini-Uber with a medallion, they don’t necessarily take care of the drivers or the cars.
The degree of techness of a company is a function of how long it can survive if you start firing their IT team.
"Tech" is simply a company that you can convince VC's to throw money at in the hopes of unicorn valuations in 18 months.
Because we all know that VC's don't invest in non-tech companies anymore.
Where I'm from, the government bought back the taxi licenses from taxi drivers and changed the law so now Uber doesn't have an advantage over traditional taxi companies.
Do they even appropriately license now? Gig economy is like Napster but with actual victims.
Can't speak for Uber... but yes. Just having an app does not make you a tech company. Tech companies tend to build stuff.
They are a logistics and pricing company. If they would license this tech they'd be rich beyond belief. They companies could really be Uber-for-X.
That's an interesting thought experiment.
At the high level, Amazon is also primarily a logistics and pricing company.
They could also license their tech (or white label it a la Shopify) and make a lot of money, but they realized at some point that owning the user relationship was key to unlocking more value than just owning the tech.
I think Amazon is licensing that tech to all 3rd party sellers now right?
Most companies prefer logistics/pricing models to be profitable. Being a pump-and-dump for VCs isn't something they can license out.
Isn't that just tuning compared to the race as a loss leader for market share?
No. This isn't a software company.
Uber isnt a software company?
I'm not sure they'd make as much as owning the network end to end. Vertical integration is hot right now, see Apple. Uber already has delivery via freight as well as getting into air delivery, so why would they need to license it? It just makes more sense to keep the tech proprietary and use it for their various enterprises.
It's tough to say, I think profitability would come faster than being a taxi company with an app (disregarding freight and air for now). But I think there is a much wider prospective customer base and recurring revenue licensing the tech stack as a SaaS. And that's not to say they have to give up Uber as a taxi company while they do both. It's a much more reasonable moonshot than the autonomous driving experiment they did that killed someone moving fast and breaking things. I see a lot of value there.
So Uber is throwing in $6.2B in ATG equity plus $400M in cash and getting $2.6B in Aurora equity? Sounds like a massively down round.
Helluva down round (the $400M in cash is the most telling in my opinion and reminds me of the kickback that IBM had to pay to "sell" its chip foundries [1]). If Uber's new stake is worth $2.5B, then the equity is worth only ~$2B. Last year Toyota+Denso+SoftBank paid $1BN for 13% of ATG, so by the new valuation their stake is now worth only about about $280M, or an over 70% loss in value in just one year.
[1] https://www.extremetech.com/computing/192430-ibm-dumps-chip-....
I think you hit the nail on the head regarding the $400 million in cash. This was a payment to get ATG off Uber's books, and given how much money ATG loses sounds like it was basically a "we'll pay for a year or so worth of expenses" kind of deal.
My biggest question about this deal is that is combining a 600 person company with a 1200 person company (obviously many/most of those people will not be sticking around). Those types of large technology merges are extremely challenging in the best of times, but I'm curious how it will work out in an era where so many people are teleworking (but I don't know, could even be better, as there is less opportunity for offhand office gossip).
Except, is ATG equity actually worth real money?
Seems to me this is a clear attempt to get ATG off the books (16% of net loss in Q3 was due to ATG https://investor.uber.com/news-events/news/press-release-det...). If Uber felt that ATG would actually deliver eventually, I doubt they'd be doing this deal to rid themselves of the department.
wmf was referring to last year's valuation of Uber's stake in ATG, hence "down round." Uber ATG was reportedly worth $7.5 billion April 2019 (giving Uber an ~$6 billion stake).
[https://www.businessinsider.com/uber-atg-raises-1-billion-fo...]
A few days ago I heard the phrase "I'll trade you my dead horse for your dead cow". I believe it was coined during the 80s S&L scandal, but it might be apt here.
Yeah this looks terrible for $UBER.
Why did you write this as ATG equity? Wouldn't it still be their class A/B shares? Or is it structured as a separate entity?
The article says ATG is a separate company. I'm not very knowledgeable about this kind of stuff so I could be wrong.
Genuinely wondering -
- How a 600 person company will determine who among the 1200 at ATG gets acquihired?
- How will the Uber employee unvested stocks get paid out (or not) in an almost no cash deal?
> How will the Uber employee unvested stocks get paid out (or not) in an almost no cash deal?
There are two types of sales that are beneficial to a companies bottom line: selling something of value for cash, and selling a liability for little to no cash.
Uber did the latter of the two. Effectively taking a business unit that was 5-10 years behind Waymo and Cruise and dumping it off on someone who might be able to turn it in to something in exchange for a piece of the future upside. But the real win for Uber shareholders here is that it isn't costing Uber R&D dollars anymore.
Why would Uber selling a portion of its business be relevant to Uber employee unvested stocks?
Do you mean Uber ATG employees? Were they getting stock packages for ATG or for Uber?
I meant the ATG employees. And they were getting Uber stock as part of their compensation. Source: me, a former Uber ATG employee.
Right, so if you are getting Uber stock, presumably nothing happens? Or I'd assume it's the same as if you'd been fired.
This is not a liquidity event for Uber. Uber is selling a portion of its business, not being sold.
They also are selling their "flying taxi" group to Joby.
https://www.theverge.com/2020/12/2/22086597/uber-sells-flyin...
Time for Uber to actually try and make a profit off of their core business, perhaps?? Good luck!
I was approached by Uber's self driving division a few months ago and I asked about the layoffs and changes there. They said not to worry about it. Turns out I was right in the end to worry about it and not work there.
The COVID situation is negatively affecting their revenue streams, which puts pressure on them to cut losses fast. You made the right choice indeed.
Signals the end of the self driving delusion, at least for 10-20 years.
I wish! The self driving delusion is more than alive and kicking. It’s just that the hype is moving from the vertically integrated options (Waymo) to the disaggregated ‘stack’ players (Luminar, Scale, Aurora)
Perhaps you mean comma.ai and not scale.ai?
Scale sells data labeling solutions to autonomous vehicle companies
Nah, companies on the brink of collapse will keep saying cruise control is self driving. Companies with a foundation keep calling it cruise control.
All about risk.
Was Uber leading self driving? I know they were a large player, but why would all the other companies give up now?
Elon is waiting on the line 1 with some promises.
You can see lots of videos filmed by Tesla car owners (not by Tesla) showcasing exactly how the Tesla full self driving beta is doing: https://www.youtube.com/watch?v=G8IQFb3JeEo
It's not perfect, but it keeps better with every release. They are ahead of everyone else by a long shot.
As far as I can tell all the videos have a licensed driver in the front left seat. I don't know how anyone can call that "full self driving", let alone ahead of other players.
Erm, there's the word "beta" in the name. Of course you want to test it with someone in the driver's seat.
They're ahead of other players because they have:
1. Hundreds of thousands of cars on the streets with FSD hardware
2. More data than anyone else
3. Their own neural network accelerator chip.
Given this is tech site - can you please tell me there Tesla is storing this data? Video is very expensive to store and process, so if they have this massive data advantage, where are their data centers?
I’d like to remind you as well that Google, aka waymo, had their nn accelerators for half a decade already.
I don't work at Tesla, so I don't know where they have their clusters, but there are obviously many cloud providers who would be more than happy to sell you as many petabytes of storage as you would like to buy.
> I’d like to remind you as well that Google, aka Waymo, had their nn accelerators for half a decade already.
That's true, but you need something mobile and low power for a car. I don't know if Waymo really has an equivalent there.
Also, it's an advantage for Tesla to have their own regardless, because of cost. Tesla can probably make their FSD computer for $200. If you are a third party and you want to buy Waymo's solution, they will sell it to you for thousands.
It still fails at very basic driving activities - it’s L2 system, and is advertised as Tesla as such (driver needs to be ready to takeover at any moment).
Very good L2, but nowhere close to customer ready self driving.
It's definitely bordering on Level 4 at this point. If you're claiming it's L2 then you simply have no idea what Tesla has been up to.
They’re not even at level 3 yet. Level 3 allows you to take your eyes off the road, and car will notify you if there’s something it cannot deal with, in advance. Tesla FSD will kill you if you do that.
As I said, it’s very impressive for level 2. But just because it can do more stuff that any other publicly available l2 system, it doesn’t make it self driving system. And Tesla knows that. They’ve quietly walked back most of their claims. They used to advertise it as level 5, that you can go to sleep. Now they advertise FSD as system that requires full human attention, and only missing thing is auto steer on city streets.
The beta has steering on city streets, it handles roundabouts, merging, changing lanes, construction work, deers crossing the road, garbage trucks, etc. It can also handle about half of the 10-minute drives I've seen without disengagements, and when it does disengage, it's rarely in a dangerous position.
You can call that L2 if you want, but I feel that's disingenuous. It's way ahead of anything anybody else has except Waymo, with the difference that Tesla's system will drive anywhere in the US.
You can say, sure, but you still need to watch it all the time. Yeah, but the number of interventions that you need to make is becoming less and less.
And Uber stock in after-hours trading is not even flinching. Wasn't this like, THE whole strategy?
I think it was the publicly stated strategy because saying "we want to monopolize taxi service, then raise prices" doesn't sound great to regulators.
When has monopolizing a sector then jacking up the prices ever worked as a strategy?
Railroads, telecommunications, aerospace, integrated circuits, electric power...
I think a better question is when has it not worked?
Aerospace doesn’t have a monopoly except that strictly regulated industries prefer having one passenger plane manufacturer because they’re easier to watch. Nor do ICs (there’s at least four big fabs and three big CPU producers).
The ones that work own moats like land or government licenses. But Uber doesn’t have that.
That's not the reason aerospace is essentially a duopoly.
The real reason is that every single aircraft development program in recent history has been late and over budget, and now aircraft development is something that costs an eyewatering billions of dollars, if not in the tens of billions of dollars at this point. Bombardier sold, Embraer tried to, and Japan, Russia, and China have all mostly sputtered on despite having well financed government backing.
To put this in perspective, 787 development cost $32B. 737MAX is expected to cost past $18B. You can't bootstrap a startup if it takes tens of billions of dollars to get a plane out of the factory.
I don't know - how many tens of billions did VCs put into Uber or WeWork, when these companies are not making a dime...
It seems like that's not a factor. Perhaps they find it boring, or perhaps the technical risk is just too high. It's all or nothing. You can BS your way through a classic tech startup as Adam Neumann did. It's harder with something that is a single flagship product that has to drive (Nikola) or fly hundreds of people in the air.
They may not be making a profit, but they are booking some revenues, in the sense that people are paying money, right now, for rideshares.
This is not really a dynamic in aircraft programs, where delivery of aircraft takes years.
It worked for WalMart.
What does walmart have a monopoly on? Are they no longer cheaply priced?
makes sense.. they can do that now with the new bill they got passed after massive campaign
It's official today, but the rumor of Uber selling ATG was already known: https://techcrunch.com/2020/11/13/uber-in-talks-to-sell-atg-...
Sounds a lot more like a merger than sale:
> Uber is handing over its equity in ATG and investing $400 million into Aurora, which will give it a 26% stake in the combined company
what happens to Uber now when Waymo releases self-driving taxi? Will they lose all competitive advantage?
Assuming that ever happens, they will likely just license it from Waymo. I don't think Waymo intends to run a full stack taxi service, easier to just sell the self driving tech itself and let other's doing the work on real world applications.
Uber's competitive advantage now is their mindshare, app experience, and subsidized pricing. With or without self driving doesn't really change that.
Maybe. The amount of work to make a taxi service is much, much smaller than that to make a self driving car. I could see Waymo entering lots of industries like that to capture easy profits.
Technical work is relatively easy to copy. But what Uber has and what the market reliably undervalues is how expensive it is to acquire customers. Furthermore they have an actually competent customer support team to deal with problems encountered while providing the service.
if you really have self-driving taxi and it is cheap, I don't think it is difficult to acquire customers, since customers are not loyal to taxi brands, they are loyal to price
It’s quite a delicate balance between supply (cars) and demand (riders). If you have too much supply, you aren’t making money because you have to subsidize all the cars without riders in them.
If you have too much demand and not enough supply, then wait times shoot up, and people return to your competitors.
Plus demand is non-uniformly distributed across time. If you have enough cars to handle peak demand, then they’ll be sitting idle the remainder of the day/week.
Most people think that trains are busy, because most people ride trains in high-demand times of year. If the same is true of a TNC, then it’s much better to list your cars with a bigger network and let them keep your cars busy more than you can.
It’s much more expensive to properly balance the marketplace than you think.
I dunno, the profits are easier and risks lower to just license the tech. Sticking to your core copmentency so to speak.
when has Google been known to do that?
lol fair point.
> Uber's competitive advantage now is their mindshare, app experience, and subsidized pricing. With or without self driving doesn't really change that.
I think the last one is their big problem: Uber has their current market-position by selling below cost, and they've locked in a lot of things like compensation at the assumption they were profitable to a much more fanciful degree. Once the VC money runs out, they're going to need to raise prices and it's a lot easier for anyone else to compete with them since they're all renting the same Waymo cars and all of the drivers are running multiple apps anyway. Uber has some economy of scale benefits but they also have a reputation for mistreating drivers and a lot of expensive real-estate and staff which local competitors don't have.
They won't. Unless that is "self-driving taxi" becomes redefined to mean "on roads with no human drivers and no pedestrians".
Check out this video, it works really well: https://www.youtube.com/watch?v=tBJ0GvsQeak ("Riding in a Self Driving Car - With Nobody Up Front")
No pedestrians, no roadside construction, no bikes. I would like to see a self driving video in major city during rush hour.
The channel has many videos with pedestrians such as in front of a Costco or busy stores, Waymos encountering weird construction setups sometimes leaving no full lanes available, and might have encountered bikes. Also, there are some during busy times instead of sleepy suburbs though admittedly not New York-levels of crazy.
And in talks to sell air taxi service. Uber Elevate tested helicopter flights to JFK with limited service. It seemed like they were betting big on eVTOL.
"Uber sells" "Uber takes stake"
I....cannot put +/- signs on the balance sheet for this transaction.
Is Lyft going to follow suit with Lyft L5?
If they can get a better deal than Uber did. Almost certainly.
At this point, any deal would be good for them
With Prop 22 in place the drivers are robots now.
Uber seems to be selling bits and pieces of itself every year. Perhaps it is wise to raise cash and focus on its core business.
When will investors learn to separate hype from what's possible - Augmented reality, blockchain, self-driving cars, AI? The reality on the ground is a lot different than all these startup peddlers have you believe. This makes it a lot difficult for people who are actually building useful things raise money.
Nice move for everyone around
F for Uber’s optics but it is clear how this is good for them nearterm
Ha! I have been saying this for at least 2 years. I see no reason for them to be in the self driving space. That is not their forte. In fact not burning cash there will make them more flexible.
how is this a sale, exactly? uber seems to be giving away their self driving division and $400 for the right to be a privileged Aurora customer in the future.
I'd be shocked if Uber is around in the next 5 years.
I'm not sure why they wouldn't necessarily be around. They increasingly raise prices to sustainable levels and maintain some share of the gig economy taxi and delivery business with pretty good brand recognition. Becomes a much more mature, lower margin, steady-state business--presumably with a whole lot less SV engineering.
> much more mature, lower margin, steady-state business--presumably with a whole lot less SV engineering.
sure reminds me of a taxi company (without the fleet management of course)
Taxis have been around for thousands of years. Taxis that you get with a nice app seem like a thing that will always exist.
Taxis haven't been losing billions a quarter for thousands of years...
Why would they need Uber? There are hundreds of similar taxi apps
Hundreds? With Uber there's a car maybe 5-10 minutes away at all times, that's certainly not true with most of their competition.
Drivers are not exclusive to uber, no? There is no moat in this industry, and its going to face inevitability when it runs out of money.
once self driving arrives, the network effect is a lot less prominent, since the supply side can more easily be scaled
And when are self driving cars going to start driving themselves around and pick people up? Certainly not within 5 years.
Isn't it already happening in Arizona?
OK, giving Uber 5 more years is pessimistic, but whenever self driving does arrive, what will Uber's value prop be?
We’re in year 35 of self driving cars being a few years away.
Once you go public, you become 5x harder to kill than your average company, and US investors have shown am increased appetite to fund unprofitable companies in the last few decades, so Uber will be around for a while
Time to buy shorts then.
Another nail in the coffin of "self driving cars will be here very soon!"...
I don't understand the current state of self driving cars.
First we have tesla which already has a self driving car. They were the first to build one and now everyone is playing catch up, or are they?
Waymo has been building something for a really long time too but I have no idea how close they are to an actual product.
Now there is this new company that gets a bunch of money to build something.
It feels to me like the self driving car market will go to the first player who can make a self driving car -- which is tesla.
Or maybe no one is close to making a really good self driving car yet, and even if they did, they would have to convince people like me that I can sleep at the wheel while surrounded by massholes.
As someone who worked in the industry, there are a couple of different approaches, incredible amounts of hype, and disappointing amounts of progress to fully autonomous vehicles.
* Tesla driver assistance features that they market as self-driving features. You need a driver that is alert and monitoring the vehicle because it can return control to the driver without warning.
* Waymo has launched limited L4 pilot in the "easy" Tempe, AZ area. They have remote teleoperation / monitoring but no driving in the car.
* ATG has prototypes that run with safety drivers
* Cruise has prototypes that run with safety drivers + they are working on tele-monitoring / teleoperation.
* Comma.AI has a consumer-available L2 driver-assistance device that you can add to your car to do ACC and lane keeping.
* Aurora has prototypes that run with safety drivers
* Zoox (bought by AMZN) has prototypes that run with safety drivers
* Smattering of other companies that are working on the tech, but are no where close to a product launch (Lyft, Toyota, Argo etc)
In general, there was a bunch of hype and bunch of money poured into the industry in 2016-2018. Others saw people pouring money into it and assumed a breakthrough was close and so they poured more money in. The cycle continued to spiral into huge multi-billion dollar investments into technology that is still IMO quite far away from a public launch. Turns out that several years later, it's really hard and no one is actually that close to useful AVs.
Waymo has a publicly available product launched, but I'm confident they are running that program at a significant loss.
Cruise has struggled to hit the milestones set in their funding deals to unlock the next levels of funding.
In my opinion, Comma is going to crush it in the next 2 years, Tesla will keep refining their AP tech, but still won't be FSD. I think autonomous robotaxis will be the norm eventually but not until after 2025.
> In my opinion, Comma is going to crush it in the next 2 years, Tesla will keep refining their AP tech, but still won't be FSD. I think autonomous robotaxis will be the norm eventually but not until after 2025.
Curious why you think Comma is going to crush it vs all the other companies working in this space?
I think Tesla and Comma are by far the best positioned for the consumer tech. Trucking is not something I feel qualified to comment on, because I don’t have a sense for the economics. But I know a bunch of R&D is going on here.
Most players in the space made investments in driver replacement tech which is multiple years from profitability. I think the places that this kind of tech makes sense is in licensing it to TNCs or operating a fleet of vehicles that you list on a TNC’s network.
On the other hand, driver assistance features are the only systems that can generate meaningful revenue today. One of Tesla’s finest selling points is their AP software.
In this vein, Comma is the Android to Tesla’s tightly integrated iOS. They are profitable today and ultimately have a larger TAM (new cars vs Tesla owners).
> it's really hard and no one is actually that close to useful AVs
My suspicion is that the thing everyone's going to founder on is that a large part of safely operating a motor vehicle on the public roadways consists of observing and then predicting the behavior of all the drivers, cyclists and pedestrians. So you need to either have true AI, or somehow get all the humans on those roadways to stop behaving so erratically. The problem is, one of those is currently impossible and the other is currently infeasible.
(I'm not sure which is which, and I'm not sure it matters.)
Are you concerned that lack of progress could lead to a sort of "self-driving winter," like the AI winters experienced in the past?
> First we have tesla which already has a self driving car. They were the first to build one and now everyone is playing catch up, or are they?
They have a Level 2 system, which is certainly an impressive technical achievement, but not what most of us are thinking of when we think, "self driving." They are hoping to grow it into something more, but it remains to be see whether that will be possible. Other companies also have Level 2 systems, and, while I've personally never used any of them, some sources do report that Autopilot is not the best one out there, just the most hyped one. So, even if we are just focusing on assisted driving rather than self driving, I'm not sure one can say with too much confidence that they are the clear leader.
I would not interpret the existence of a functional consumer product as evidence of a technical lead in the race toward true self driving. Waymo never had any intention of commercially releasing a level 2 system. Their plan was to go straight to level 4 without any detours. Reports would seem to indicate that they are far, far ahead of everyone else on achieving that goal. It may be that they now have an unbeatable lead. Or it may be that some more direct competitor catches up. Or it may be that a company like Tesla or Cadillac is able to use profits and data collected from their commercial level 2 systems to accelerate their development and catch up. This, too, remains to be seen.
My intuition is that Waymo and Tesla dominate in the amount of data available to them.
Waymo seems to be closest to a robust 3D lidar based solution, but Tesla maybe be closest to 'a' solution, but the community is split on whether their 2D approach will work.
Then there seem to be a laundry list of companies that entered the market late, and are propped up panicking car companies which aren't sure which one of these companies is the real thing. So, they hedge their bets by taking a piece of all of them.
The solution will come out of 1 of the following:
1. A slow race to making a more and more robust vision system that is eventually good enough to get through regulation. In this case, the earliest to start and one with the most data will win. So, either Waymo or Tesla. By the time others get there, they will have captured the market and made billions.
2. Some smaller company has an 'a-ha' moment. Waiting for an a-ha moment doesn't sound like the most prudent investment decision, but the one that finds it will become a billionaire overnight. This change is so important (like Mapreduce) that it will quickly be reverse engineered and one of Google/Uber/etc. will release a self driving hardware kit + api or come pre-installed in cars like android auto. The solution transforms the vision field with a reach far beyond just cars.
Put down the kool aid. There are no self-driving cars.
To be fair, the levels of self-driving are almost never discussed so it's no wonder people say things like "Tesla already has self-driving cars".
For those unaware:
- Level 0: No automation
- Level 1: Driver assistance (Assisting steering or braking)
- Level 2: Partial automation (Assisting steering AND braking)
- Level 3: Conditional automation (Human supervised autonomy)
- Level 4: High automation (Full automation in ideal environments)
- Level 5: Full automation
I believe we're currently at Level 2-3.
More details here: https://www.howtogeek.com/401759/WHAT-ARE-THE-DIFFERENT-SELF...
If someone with more knowledge wants to correct me on the levels, feel free.
There's nothing "to be fair" about this.
The levels are technical distinctions in a roadmap towards "self driving cars", which by any reasonable definition are Level 4 and 5.
Below that it is marketing for advanced cruise control features (level 1 and 2), or dangerous (level 3).
To be fair, I started with "I don't understand the current state of self driving cars".
I own a Tesla with FSD. Can confirm, there is no self driving car.
> First we have tesla which already has a self driving car.
What 'self driving car'? I think you meant 'semi-autonomous cruise control' as there is no Tesla car that is fully 'self driving'.
As for Uber's situation, they had no chance. Mounted losses everywhere caused by the pandemic, government lockdowns and drivers who are now classified as Uber employees make the costs of it all unsustainable for them to justify keeping a 'self-driving unit'.
The only certain and sustainable long term players in this 'self-driving' or 'cruise control' game are Tesla and Comma.ai.
I was fascinated to learn that Comma are the only (?) company whose ML algo dynamically segments the road and surroundings, whereas I believe Tesla specifically looks for things like lines on the road.
[https://medium.com/unikie/self-driving-cars-2020-and-beyond-...] helped me understand the major players in self driving vehicles and what they're upto.
Tesla does not use Lidar as a cost savings measure, and only uses camera output to determine how the car drives. It is unknown if this is enough to account for enough permutations of traffic, but they're calling it full self driving while using car owners as paying (!) beta testers.
AFAIK, no one is close to releasing an autonomous vehicle that can drive on par or better than an average human driver, but that several have a path to getting there.
I believe many companies can thrive until M&A whittles down that list. I don't think this is a first mover takes all kinda deal because safety will have to be proven over a long time allowing other entrants to prove their worth.
>Tesla does not use Lidar as a cost savings measure
I think it is less about cost saving and more about getting to the root of the problem. According to Elon's statements, Lidar is a crutch and since the AI will eventually have to be good enough to remove the crutch, remove it an the get go and save learning how to use the crutch. This is similar to his reasoning about EVs. Since we will eventually have to have petroleum free cars, Lets remove them as early as possible.
Its a cost savings measure. Tesla is adding an imaging radar to their cars (which gives a LIDAR-like output) because their current sensor suite doesn't cut it. "LIDAR is a crutch" is Elon trying to rationalize why he advertised cars without the necessary hardware as capable of autonomy.
(See https://mobile.twitter.com/greentheonly/status/1319363515610...)
Sometimes you need a crutch because your legs are not up to the task of walking. And the only way to recovery is to use them
> According to Elon's statements, Lidar is a crutch and since the AI will eventually have to be good enough to remove the crutch, remove it an the get go and save learning how to use the crutch.
There's zero reason to believe him on this.
LIDAR - combined with vision - gives a much better awareness. The only reason not to use it is (or was) cost.
> First we have tesla which already has a self driving car
No, they don't. And it will take a while to get there.
I think there are multiple companies that can make something that works most of the time in good conditions, but you have to have a certain level of arrogance and self confidence to be the first to release it as a product.