Analysts put GM and Waymo far ahead in driverless car race | Hacker News Reader

Analysts put GM and Waymo far ahead in driverless car race

arstechnica.com

61 pointsBerislavLopac75 comments

What is also interesting here is GM's Cruise doesn't need Uber to do what Uber does.

However Uber needs cars to do what they do.

I have always argued with my friends that ai cars are more likely to kill uber than help as car manufacturers would start their own taxi service if the ai they have is not good enough would partner with waymo to stay in the market. Uber brings no value.

Uber's value nowadays is that they have the infrastructure as a "connected" taxi service, and they have the users. But I guess that's not unique, Lyft has this too, and installing another taxi app or two is easy enough (or in the case of Google, they'll probably just integrate it to Google Maps -- I just realized why they've integrated Uber into Google Maps, so when they do that with Waymotaxi, no one can scream "anti-competitive move!").

Finding and vetting drivers will be the slightly harder task...

I just realized why they've integrated Uber into Google Maps

Google Ventures invested a large amount in Uber as well – I think they owned something like 5% last time I looked.

> I just realized why they've integrated Uber into Google Maps, so when they do that with Waymotaxi, no one can scream "anti-competitive move!

No, they integrated it so that Maps becomes the go-to point for users when they need a ride. That way, Google can snatch those users away when Waymo becomes available. The internet nowadays is all about being the "gatekeeper", and Google just wants to stay in that role.

Google will do the smart thing too. When Waymotaxi becomes available, they won't remove Uber as an option because that might be anti-competitive. But more relevantly, they wouldn't need to. Back of the envelope calculations say that a self driving cab would be 1/3rd of the cost. Show both in Google maps and let the people decide.

> Back of the envelope calculations say that a self driving cab would be 1/3rd of the cost.

I can hardly imagine taking the human drivers out of the equation will be that profitable. Right now, those drivers also do the labor intensive part of maintaining the fleet (gas/power, cleaning, repairs etc.). That part will stay with uber, lyft, google etc.

When car companies will become taxis, repairs will stop being profit centres, so the incentive will be to make more durable cars. Also electric cars will become more convenient with AI Taxi. Uber drivers don't use their cars 24/7. AI Taxi can deliver good between companies in the night. There is so much room for improvement. Right now it won't be 1/3 of the cost, but in the future AI Taxi will cost as much as public transport.

> When car companies will become taxis, repairs will stop being profit centres, so the incentive will be to make more durable cars.

I think manufacturing cost is a much stronger driver of shitty reliability than repairs being a profit center. After all, in the US dealer system the manufacturer isn't getting the profits, the dealer is.

> Uber drivers don't use their cars 24/7. AI Taxi can deliver good between companies in the night.

Keeping complex systems like that running constantly is difficult and expensive. It also wears said systems out faster and increases capital costs.

> the manufacturer isn't getting the profits, the dealer is

Pardon my ignorance, but why doesn't the manufacturer open a dealership then?

This maybe US specific. Google "tesla can't sell directly to customers"

It also wears said systems out faster and increases capital costs

This is a good point. When you purchase capital equipment, your incentive is to keep it utilized as much as possible to maximize ROI. That means you're going to want your self-driving car fleet parked as little as possible because when they're parked, they're not making you money. Now that the cars are driven far more than regular passenger cars, their maintenance needs go up.

So either these cars will need to be built to the reliability needs of over-the-road trucks, which will greatly increase costs, or their maintenance will have to be more frequent, which is even more expensive.

They will move to electric cars as they have fewer moving parts so maintenance compared to petrol cars will be a lot less.

There are fewer moving parts in the engine and (perhaps) transmission. Same number of moving parts everywhere else. And remember it's a car: things you don't think of as "moving parts" are still in motion and will be degraded by vibration, dirt, road salt, etc.

Yes, that's what I had in mind.

> they have the users

As you pointed out, this is a very fickle (price sensitive) group. Anecdata-ly, I can point to many who prioritize Lyft based entirely on personal price comparison history.

I’ve long thought Uber’s plan was acquisition: google up cash to GBF (Get Big Fast) with unassailable network effect (brand, user data, city usage data etc) and then sell off. Just as GM got Cruise (for a song) then a GM would buy uber.

That seemed, for the reasons you gave, to be a long shot. Sure the car mfrs don’t have services in their dna, but it won’t cost THAT much to get there. Certainly not the $40-odd billion valuation Uber bears.

I also have long believed there will only be 4-5 global car brands once the vehicles drive themselves (fleet production in a monopsony environment destroys the value of branding). I hadn’t guessed that GM might even have been in the runnng.

> there will only be 4-5 global car brands

We heard the same thing about mechanical watches when smartwatches came out.

But actually what happened was the opposite. People flocked back to mechanical, vintage watches and the watch industry is booming right now. I suspect the same thing will happen with cars. There will be even more car brands than we have today with old classics coming back e.g. Saab but all built off cheaper, commodity electric drive trains and far more differentiated.

So you will have AI cars taking over family movers, taxis, buses etc and old, classic cars being modernised and built for drivers.

Cars right now are both a practical good and a luxury item (where branding comes in).

Clocks were the same, until everyone got a phone in their pocket at which point they were reduced to just a luxury item.

I'm sure there will always be car collectors, but buying a car as a luxury item when there's no longer a practical need to own one, is far from comparable to buying a watch.

I have said it before, and will say it again: I do not understand why the assumption is that fully self-driving cars will eliminate individual car ownership. What are the practical reasons to own one now, and why will they be rendered irrelevant by AI drivers?

If I don't want to own one now, I have to get to some place where they rent out cars, get it, do my thing and get it back there again, and then back home. For me this takes about ~30 minutes door to door each direction, even though I live "in the city".

In a self driving scenario I use an app and it shows up at my door, and likely much quicker, and then it drives itself off after I'm done.

You can already do this now via Uber, Lyft, or a similar service. Other than making such a service somewhat cheaper, what fundamentally changes?

> You can already do this now via Uber, Lyft, or a similar service.

This is heavily location dependent, as smaller cities and even countries don't have these services. And as you said, much more expensive than a driverless car. Also having the car wait for me during errands would cost a lot more if there's a driver sitting in it during.

Assuming that most of the cost of taxi and rideshare service comes down to paying the driver, I'm expecting that fully self-driving cars will allow the price of such things to drop precipitously. I expect that it will easily be enough of a price drop to make taxis the cheaper option for many people.

It may also change the practicalities in ways that drastically alter the nature of the service. With no driver to worry about keeping happy, it might be perfectly fine to leave taxis just parked way out in the middle of nowhere, waiting maybe hours or even a day or two for the next ride. That makes using taxis to do things like go on a camping trip, or haul yourself out to your parents' house back on the farm, something that makes sense.

> Assuming that most of the cost of taxi and rideshare service comes down to paying the driver

From the data I have seen from the New York Taxi and Limousine Commission, this is not a good assumption.

> I'm expecting that fully self-driving cars will allow the price of such things to drop precipitously. I expect that it will easily be enough of a price drop to make taxis the cheaper option for many people.

While self-driving capability will definitely make taxi service cheaper (even if I think you overestimate the size of the drop), cost is not the only driver. I would not switch entirely to such a service because I want the convenience of having a vehicle available as soon as I want or need it, and the ability to leave things in the vehicle for extended periods or just in case I need them. The rest of my family would have similar issues.

> It may also change the practicalities in ways that drastically alter the nature of the service. With no driver to worry about keeping happy, it might be perfectly fine to leave taxis just parked way out in the middle of nowhere, waiting maybe hours or even a day or two for the next ride. That makes using taxis to do things like go on a camping trip, or haul yourself out to your parents' house back on the farm, something that makes sense.

I don't see how this is any different than a car rental service now, other than removing the pick-up and drop-off portions. So I still don't see how self-driving capability fundamentally changes anything.

Can you explain where else the cost comes from?

I'm thinking here that if a taxi driver's earning, say, $35,000 per year, and you can have two drivers working a shift per car, per day, then your labor costs for operating a cab are $70,000 year per car. Assume a car drives ~100k miles per year, gets 25mpg, and gas costs $4/gal. That's $16,000 for gas. How much do the cars cost, and how much maintenance do they need?

The problem is that everyone who answers this question focuses on price while completely ignoring latency.

Every morning at 6:30, I drive my son to the school bus stop about half a mile away. I do this in part to make sure he makes it there on time, and also because it's often well below 0F at that time of day. So, even if it costs $1 to get an Uber to do that, it would have to show up within a 3 minute window (time between when his morning chores are done and we have to leave), and it would have to wait 5-10 minutes for the bus to show up, since he doesn't want to stand around when it's -10F and blowing 20+ MPH. That's not a normal behavior for a taxi-type service.

This daily trip and my morning/evening commutes are the only scheduled vehicle uses we have. Every other use is unplanned and random. The nearest grocery store is 10 miles away. Now I have to plan my trips or wait for a car to get here, take me there, and then wait for another one to pick me up? Oh, I forgot milk. Call another Uber to make the 20 mile round trip again?

I can see this behavior being OK with people who don't have cars to begin with, or who live in very dense areas, but for the outer suburbs and rural areas and other places built with the assumption that people are driving cars, it's really not workable unless people's expectations drop considerably.

>it might be perfectly fine to leave taxis just parked way out in the middle of nowhere

Where are they going to park and who's paying for that parking lot? You can't just handwave these things away.

It’s the largest, or second largest (if you have a mortgage) investment people make, and despite that it typically has less than 5% utilization rate. We’re not talking about cake TV channels — making car payments is a major source of heartache for a lot of people, Eeven worse than making rent since you typically need a vehicle to earn your rent.

Large companies can borrow for less AND get a much higher utilization rate from the asset.

I just ditched one of my cars. I had no loan, but dividing the purchase price by the 8 years I’d had it, plus insurance and registration it cost me ove $15/day just ito have it in the driveway. I spend less than that on Lyft/uber (though I have a pass for the latter)

For now. Imagine that many people give up their personally-owed cars and rely on subscriptions in the next few years.

Sales of cars would drop fast for manufacturers who don't have a on-demand service network with a sufficient critical mass.

Don't forget that giants went close to bankruptcy around 2008 with only a few % of global sales decrease.

Their market cap could drop low enough for Uber/Google/Apple (and their investors) to acquire them on the cheap.

Uber has massively subsidized users for years, but they don't have the cash to keep doing this for very long or buy a fleet of self driving cars. When self driving cars do take off you can expect a spike of sales as people really want that feature. So, Uber would need to survive ~15 years before any major car companies feel the squeeze and I doubt they can make another 5.

Subsidizing users is one side of the coin. The other side is that Uber has been living beyond their means and they do not have the cash to keep doing this.

Uber would only need to raise fares 10% and not increase driver payouts to be in the black. The dire predictions don’t seem connected to the balance sheet I’ve looked at. Everybody wags their tongue at uber quarterly “loss” but they pay no attention to the more important measure of free cash flow

Break even does not enable them to pay for massive investments. A 20% increase would cost them significant ridership, but is more or less required to have any significant profit.

They would finance it through debt and equity, do it post IPO etc. There could be a merger instead of an acquisition. Currently if they cut their R&D by a bit and raise prices by 10% their more than healthy enough.

If they only raise prices by 10% it will cost them rides and drivers to Lyft / taxi etc. It's hard to say how much market share this would cost them, but you can't assume their numbers are going to be static with a price increase or they would have already increased prices.

10% is plenty to drive business to Lyft, especially with Lyft’s better brand.

The part about bankruptcy over a small dip is actually really interesting.

That probably means that a pretty small lead in this area could be very significant in terms of how this actually plays out.

Weren’t there other issues? Pension debt and union issues were hurting those companies for years, the slowdown just nudged them over and then a lot of that stuff got restructured.

There are a ton of variables, not having a self driving play seems like a concern for an auto company, other than that it seems like it could play a lot of ways and there is still this giant culture thing where cars symbolize so much

That is partially happening already. Even Porsche is offering a subscription. [1]

[1] https://techcrunch.com/2017/10/10/porsche-launches-on-demand...

I don't think this will ever be worth it for users that need a reliable ride at any time (e.g. families). These new subscription services are great if you want to drive a new car all the time and experience different models. But if you want to use the car whenever you want with no hassle and without cleaning it after/before every use, ownership (or leasing) is still the way to go.

GM has a massive distribution, storage, and service network that Uber will have to replace if they are going to stop relying on drivers buying, storing, and maintaining their personal cars.

This absolutely will be a big shift for the existing manufacturers, likely resulting in major cuts in manufacturing employment.

The question is: are they taking steps to be prepared for the change, or will the change catch them off guard?

It seems like GM is preparing itself, at least on the tech side. Is it taking similar measures to bolster and prepare its balance sheet? That’s unclear to me. But that’s what I’d be looking for.

I don’t disagree with your comment, but wish to add one important note:

Once cars are pure electric (not hybrid), then that maintenance network becomes fairly worthless.

I think Tesla has shown how little maintenance an electric car needs, and how much smaller the related infrastructure becomes.

Any car still has suspension, brakes, electronics, body, tires and all the small gimmicks .. And autonomous cars will bring on even more sensors and electronics. Existing maintenance network already deals with sensors and electronics.

No no no you see the way this works is let's say the car you are in gets a chip in the windshield from a rock... It will immediately dial home for a solution which will most likely be to drive itself off a bridge to get a total replacement, and forge the records by generating a video of a non-driverless car swerving into it, with a database of license plates of known driver based cars..

Uber is an app and some fast talking bros (to put it a bit loosely and little harshly).

GM actually builds things. There is a huge difference.

Point being, it's probably going to be easier to to build the app and the rental network than to start building self driving cars.

Great point.

I’d love to see GM create their own app, and offers drivers who use their app discounted pricing or financing on their cars.

Could create a whole new world of competition for Uber.

If GM is really only 2-3 years away from having a fully automated taxi service, I don't see the strategic value in making such a move.

A whole bunch of time and money sunk into launching a rideshare service and recruiting drivers, only to almost immediately transmute the entire thing into a giant PR disaster full of news stories to the effect of, "A year ago I mortgaged my house to buy a car and drive for GM, and already they've kicked me to the curb."

I’m presuming that it’s longer than 2-3 years to truly get full automated driving.

2020 is still the goal for geofenced city taxi autonomy for Uber and Cruise.

It may be, but I don’t buy it. Of course, I’ve been wrong before.

I don't buy it, either. But my original comment was speculating about GM's perspective, not mine.

German manufacturers already own uber rivals. The reason not more car companies have them is because it doesn't make sense financially and there's no lock-in. You can get an app up and running within a few months and users will be happy to switch if it's cheaper and/or more convenient.

If GM manages to build self-driving cars before Uber, why wouldn't they just acquire a third tier Uber competitor (someone small enough that they don't have to pay for tons of existing users, but big enough to have made a reasonably functional app), kick out the human drivers replacing them entirely with self-driving cars, then undercut Uber's prices driving Uber out of business?

It would be much, much easier to get close enough to Uber's app functionality to compete with them than to get close enough to a self-driving car to be able to use them for millions of riders.

Well, golly. Google is rumored to spend over a billion yearly to maintain their maps which are absolutely vital to a self driving car. That's a moat.

As for where the technology stands, it's no surprise they want to launch in Phoenix first -- it doesn't rain there...

Any candidate acquisitions for Waymo to buy the manufacturing capacity that it needs? I don't know the car industry well, but looking at the market cap of companies[1] it looks like you're talking at least $40-50B for an established manufacturer (Nissan, Tesla are in that range). That is affordable for Alphabet (~$80B cash on hand), but I'd be surprised if they went all-in like that.

[1]: http://markets.on.nytimes.com/research/markets/usmarkets/ind...

I've been wondering for a while where Waymo is getting all of its money since I've never heard them mentioned except when they're suing Uber or an article saying "Waymo's tech is awesome!", and I didn't realize until right this moment that it's a Google spin-off. That makes a lot more sense.

FCA (Fiat, Chrysler, Dodge, Jeep, Maserati, Alfa Romeo) wants to sell itself or part of itself, but I can't imagine that Google would want to buy them.

Their share price/market cap has been going crazy lately, for no apparent reason, perhaps indicating a buyer.

https://ycharts.com/companies/FCAU/market_cap

Waymo is already partnered with Fiat-Chysler -- the current test platform is the Chrysler Pacifica. Waymo also opened an office in Livonia, MI to be near the University of Michigan MCity testing grounds and the Auburn Hills, MI headquarters of their partner manufacturer.

That's a good fit then, thanks.

With the tax bill's forced repatriation, I imagine Google will be looking for something to spend tens of billions on, so that could actually be a possibility.

Good! Competition is good. JVM was just-okay until .NET came along and forced it to become great. I have my own car still because I don't like getting in strange cars all the time. Most of time time it's okay, sometimes I am with company and want privacy. Also when I need reliability it's worth paying $20 for parking in FiDi/SoMa (sometimes cheaper than a round trip from the Sunset), and it's nice for group adventures to not have a stranger in the car... slightly more/less convenient than rental options like Zip, depending on personal factors.

Driverless taxi would make me a lot less likely to own I think, covers a few more use cases. I still like cruising up Great Highway and through the park (and the whole BA is beautiful), it'd be nice if they had a leisure mode for the Sunday Drive, maybe even with parking to leave stuff while you venture away.

Link to report: http://fordmediacenter.nl/wp-content/uploads/2017/04/LB-AV-1...

This will get to consumers through ride sharing services well before buying a car with the capabilities. Transportation will just become more of a service.

The last numbers I have seen have Google still well ahead of everyone.

https://www.wired.com/2017/02/california-dmv-autonomous-car-... Autonomous Vehicles Makers Report Disengagement Numbers in ...

An aircraft carrier can't turn on a dime but once it turns you have a f-ing aircraft carrier.

When GM or some other BigCo throws its weight behind something that something gets done.

They can't innovate but once other people innovate they can copy and buy up their technology and commercialize it more successfully.

With self-driving cars the tech is only phase one. The legal/regulatory/social issues come next and that is where a company like GM with all of its political connections can really press their advantage.

Even a company like Google is small time compared to GM when it comes to lobbying and political power with all of the unions, dealers, suppliers, trade associations, etc they have built up over a 100 years.

GM has been innovating just fine in the EV segment, with the Volt, Spark EV, and now the Bolt. All three were extremely well engineered and relatively affordable -- if not as 'hip' as a Tesla. And designed mostly in house (the Bolt owes a lot to LG)

I'm fairly impressed. I'm sure like most BigCorps it's not a fun or terrible nimble place to work, but it seems like they've got at least some aspects of their engineering culture figured out.

I'm surprised and a little disappointed that Ford hasn't done more. They have the EV version of the Focus but AFAIK it's only sold in California.

Edit: 2018MY Focus Electric is a limited production model sold in select states (CA/CT/MA/MD/ME/NJ/NY/OR/RI/VT) at Ford EV Certified Dealers

https://www.ford.com/cars/focus/models/focus-electric/?intcm...

Especially considering how quick they hopped on the hybrid concept, making hybrid versions of quite a few of their cars and SUVs, quite early on.

Yes, they have been very tepid (only selling in CARB type states) and actually the quality of their EVs is subpar -- no active battery temperature management, low range.

They're making noises about improving, but then they go and do the opposite.

> Volt

Toyota Prius, a decade later.

> Bolt

A Prius with batteries. Oh my, I can't take all this innovation.

> relatively affordable

Because the billions in losses are paid for by other GM customers and taxpayers.

If not for massive government bailouts GM and other "BigCo" companies would likely already be dead. In 2009 the US government dumped $50 billion into GM alone just to keep it afloat, taking a $10 billion loss when divesting.

The self driving battle is going to be complex, unprecedented, unpredictable, rapid, and probably dirty given the stakes. These are not generally the sort of conditions where old behemoths tend to shine.

GM/Cruise developer here. GM's self-driving tech is a joke compared to Waymo's. Don't be fooled by the advertising.

So, combining two expertise - first in form of self driving software and second in form of hardware - is better than trying to build it alone like Tesla or Apple.

This same exact article was posted 20 hours ago. (https://news.ycombinator.com/item?id=16171625)

I am starting to get the impression that these websites, try and time when to push articles to HN in order to bring in visitors to their site...

Some even change the title to get a more engaging tile.

Overwhelmingly it's just regular users posting things they found interesting. Reposts are ok if a story is good and it hasn't gotten attention yet: https://news.ycombinator.com/newsfaq.html.

What you're talking about exists, but as far as we know it has little impact on the site.

Eh. Cadillac's Supercruise™ technology doesn't seem to be that advanced, really...

Reply on news.ycombinator.com