Would it be too cynical to assume this whole thing is going to get shutdown after a few years due to failing to gain enough traction (due to people being wary of Google's track record of doing so with their in-google properties)?
Also, they seem to have the lead over most and the market opportunity is huge.
The fact remains they are out there, and you can buy them, and they can do some sort of self-driving. Waymo is an impressive tech demo at this point.
Think about how much capital will be required to put a million Waymo vehicles on the road. How many zeros are there on that number? Who is going to be paying that bill?
Given this, even if Waymo are first to driving 100% perfectly, I expect that Tesla will eventually leapfrog them.
Waymo can just license the tech and sensor suite to manufacturers and make money off that. They don't need to actually build and sell a million Waymo vehicles.
The beauty of the Tesla model is that they've already made tonnes of the cars that should eventually be able to do this. Furthermore, instead of each vehicles being a cost for them, the vehicles are a source of revenue.
I acknowledge that it does all hinge upon Tesla reaching a point with their autonomy where it can be unsupervised. Some people say they will never get there, but I think that's pretty shortsighted. The data collected by their existing fleet is a massive advantage. Let's not forget that it's been more than 11 years since the first video of what became Waymo, and they can pretty much only drive a small area in Phoenix, with high likelihood of needing to train extensively for every new location as they're now doing in California. Tesla are doing most of the world in (more or less) one go.
The problem with Tesla is that while they've sold cars with that as marketing claim, there is very little reason to believe that its true.
And cars don't have an infinite shelf life; sure, they've sold lots of vehicles with that claim, but even if it were true, a whole bunch of those are already 1/4 of the way through an average US vehicle lifespan, and Tesla may end up with lower lifespan given its rock-bottom initial quality.
But where did you get a million cars from? To equal Tesla's existing fleet?
Uber has 3 million active drivers globally. If you assume that they're working in 8 hour shifts, 1 million Waymo cars is a rollout sized to replace Uber.
Coming at the math from another direction, Uber does 1.6 billion rides a quarter. A million cars would mean 18 rides per car per day. Totally doable.
But Waymo would surely focus on countries with high wages - US, Canada, UK, Australia, most of Western Europe. Replacing Uber in only those countries would only need perhaps 500,000 cars.
Of course they're going to ramp up towards those numbers, but once the tech works and it's just a matter of cash, it's going to happen very quickly.
If Waymo can get a 60% gross margin instead of a 20% gross margin, their profits could be an order of magnitude higher.
This is a trillion dollar opportunity, and its the first thing Alphabet is seriously pursuing that can rival search.
Aftermarket conversion kits are a rounding error and there's no long term moat. And that's before you start to get in to issues like improper installation and use, aftermarket modification, accidents causing bad PR when the gear isn't even running etc.
Remember that AEB demo a decade back where the car just slammed in to the object it was meant to stop for and it turned out they hadn't turned it on? Nightmare.
Yes, they definitely have the lead in outsourcing costs.
> Think about how much capital will be required to put a million Waymo vehicles on the road.
Actual deployment, not counting developing the technology, maybe, what, $100 billion or so at the outside, given a generous guess at the unit cost of the cars and a generous multiplier to cover other deployment capital costs.
> How many zeros are there on that number?
8
> Who is going to be paying that bill?
Well, I mean, if they couldn't convince outside investors, Alphabet could do it with cash on hand and still have money left over, but if they can demonstrate the technology, investors will be falling all over themselves to jump on board.
*: I don't know the actual number.
They are not even close to being able to do what Waymo does.
The question is not, who has the first taxi on any public road anywhere, but who first can launch 10000 of autonomous taxis and make money.
Waymo so far is just a gigantic money fire, and even if you assume their autonomy works, many people have made arguments that the businesses case is really not very solid at all. Waymo has incredibly few ours driven on public roads outside of a limited area.
Self driving is mostly a software problem, and to solve general driving, a gigantic learning problem. How is gone solve that first and scale it, is open.
I would put my money on Tesla, and I wouldn't invest in Waymo if I could.
Maybe if Waymo demonstrates that they can drive on and off all major highways in the world without hd mapping of them. Then i might believe they are ahead.
So, can Tesla do this right now?
Also, why should Waymo not use HD mapping when they clearly have the ability to do it? Seems like a weird criteria.
And its highly questionable that their mapping can actually deal with all the incredibly complexity of the real world.
I'm going with the better process, not the better demo.
Funny, I have the same the opinion of Tesla’s camera-only approach.
> I'm going with the better process, not the better demo.
A “demo” with 20 million autonomous miles clocked vs 12.2 miles. The “process” that has so far yielded very little in terms of actual autonomous driving if you ignore the FSD marketing speak.
Probably Waymo
> Waymo so far is just a gigantic money fire, and even if you assume their autonomy works, many people have made arguments that the businesses case is really not very solid at all. Waymo has incredibly few ours driven on public roads outside of a limited area.
and Alphabet / Google is a money printing machine, if they get to self driving first it’s pretty much over for their competition.
> many people have made arguments that the businesses case is really not very solid at all
So self driving technology isn’t a solid business model? Why are so many companies trying to do this?
What’s stopping Waymo from patenting this after they get to an acceptable threshold and selling it to their competitors?
Waymo has spend 10+ years and they can now barley driven in possibly the easiest location in the world, a location where they have logged millions of miles. Waymo focuses on the easiest thing first. The technology on their car is incredibly expensive both now and to scale it to 10k, 100k, 1M vehicles is a huge challenge, both capital and manufacturing.
They are like a rocket company who says, we have this amazing air-frame, launch pad, software we only need to figure out the engine, then we are gone make money.
I place my bets on the company that has million of cars on the road, seeing every corner case possible in every location, because that is the real challenge to overcome if you actually want to have many 1000s of cars in every city. Plus the sensor suit and compute power in a Tesla maybe cost a few 1000$, scalable to 10s of millions of vehicles.
I don't think either of them is really close to self driving, but at least Tesla is focusing on the right problem with a scalable solution.
> What’s stopping Waymo from patenting this after they get to an acceptable threshold and selling it to their competitors?
If their technology actually works they would want to dominate the self driving taxi market and not sell it to the competition.
Their competitors would then have to install this whole Waymo hardware suit, making it not very practical for OEMs and normal cars. Uber or Lyft might be costumers, but why would you sell it to them if you could just replace them.
Waymo has burned money for 10+ years, signing a few licensing deal for a slightly better Mobile Eye will not give the return they are hoping for.
They are still in a learning phase -- they charge enough so as not to distort rider behavior, but they expect to lose money on each ride.
Apart from that, why build cars when you can sell the system to everyone?
Not having lidar has directly caused a bunch of autopilot crashes, so I expect they'll add it.
Hilarious to hear this come from someone rooting for the guys who took an off the shelf LKA + ACC system that couldn't handle stationary objects, turned it on full-time, and called it "Autopilot".
Of course the decision to essentially misuse that sensor suite killed a few people but that's neither here nor there right.
I don't think you can disparage Waymo and cheerlead Tesla on the basis that you think Waymo a money fire whilst ignoring that Tesla is massively in debt and has yet to post a positive annual net income. Tesla has burnt money for 17 years and it's recent profitable quarters are only the beginning of a turnaround that is not yet anywhere near sufficient to recoup it's losses.
In the meantime legacy car manufacturers are catching up with Tesla's offering and providing ever stronger competition. The coronavirus has disrupted production & made sales numbers confusing, but Tesla has lost its top spot in markets it led until recently. Tesla has been great PR for EVs and an important early market innovator, but it is by no means guaranteed long term success in either self-driving or car production.
The difference is that Tesla had a clear buissness plan to profit that they executed step by step. And the only reason they lost money is because they were expanding production capacity.
The analysis of the cars clearly showed that they had great margin on the cars, by far the best in the industry for BEV.
The same can not be said for Waymo. Its basically burning money and not even a clear plan on how this can eventually be recaptured. Tesla already makes money with teams that work on this stuff.
> In the meantime legacy car manufacturers are catching up with Tesla's offering
Wait, are you somebody from 2014? Because we have been hearing that since then and its still not true. And even if it were, it would not be a thread for Tesla, its a thread for ICE vehicles.
> but Tesla has lost its top spot in markets it led until recently.
Tell that to the 18% global market share in terms of EV sales. Its more if we are talking BEV. And its even more if we are talking revenue. And its even higher in terms of profit.
The facts actually say that Tesla has been expanding its global market share. But of course if you only read media articles that say 'in August Tesla fell to second spot in Lithuania' and BS like that, then you end up being misinformed.
Tesla also a has 3 gigantic new factories coming online next year, plus they easily have the best strategy for battery scale.
I'll bet you $1000 Tesla won't be the first to develop self-driving that is at least as good as humans (wrt. accident rate) in all weather and terrain conditions. Hell, I don't even think they'll be the third.
Disclosure: I work at a self driving company that is neither Tesla nor Waymo.
I'm also wary because ride sharing is a customer service intensive industry, an area Alphabet isn't exactly strong at.
These were never my problems with taxi service.
Tracking is the thing that makes the rideshare services better--and taxis could apply that, if they wanted to.
Tracking solves most bad things about taxis. "Is it actually coming? How long until it gets here? Did they hit something unexpected? etc."
Don’t forget: upfront pricing. Most taxis won’t tell you the price until you get to your destination. And how do you know they took the quickest route?
Again, I can't remember ever having that problem.
If I had to phone call the taxi--I always got a price. That's fair as the price needs to be high enough to roll the taxi.
If I didn't, then the price was specified--$X from Zone A to B and $0.Y for time spent not moving.
> And how do you know they took the quickest route?
This is questionable even with ridesharing. I regularly have to fight with drivers going to my airport in the morning. "Look, I know your route says go via State 666. Don't. It's completely green now but in 10 minutes at 6:45AM that road goes from green to black AND you can't change route once you're on it--it will take 40+ minutes. Go via Interstate 13--it will take 20 minutes, tops. Everything will be fine, and, if it isn't, you can get off the Interstate and take another route, of which there are lots."
And those are exactly the kind of routes that cabbies use to inflate prices. See: Las Vegas--a good cabbie moves a couple of streets in the wrong direction to get away from the strip, moves most of the distance, and then works back to your destination.
If they take a suboptimal route, you as the rider aren't charged more.
The incentives are also better aligned, since the driver now would always prefer the quickest route, since they get the same payment regardless of the route.
> If I didn't, then the price was specified--$X from Zone A to B and $0.Y for time spent not moving.
That's...exactly the non-upfront pricing that got taxis killed. Upfront pricing is $X to your destination. It's not route and time dependent as almost all taxi pricing was.
The only question about payment is "Who?"
As much as you can fight it, at the end of the day the answer will always be "the end consumer." The only question is whether you pay it on every ride as an up front increase or whether you pay it on the unfortunate ride as an addition. Both are perfectly acceptable solutions and can even coexist in the market in different companies.
People have gotten spoiled by the fact that ridesharing was burning VC cash to subsidize exceptional cases. Once those costs all quit being subsidized, they're all going to materialize back to the end consumer.
What no one ends up paying for when incentives are aligned however, is taking the tourist from the airport on an unsolicited tour of the city rather than directly to their destination.
Or the taxi driver who accidentally ends up on that one road that always has an idiot who caused a fatality when there were others that were less busy at the time.
_That_ is what most people are worried about, not having to reimburse someone for an unforeseen circumstance.
Not to mention ridesharing apps have incentivized safer driving with their review systems much more than taxi drivers who generally felt they were beholden to no one or "their own" (which is why they felt comfortable doing things like taking people on joyrides) ever did
There's a lot of ambiguity about accident rates when a passenger is not in the vehicle with an Uber driver, but the rates for actual trips are lower than national averages: https://techcrunch.com/2019/12/05/ubers-fatal-accident-tally...
Amortizing the cost of rare negative pricing events across all trips seems a much better model to me. I'd personally choose that option every single time.
The time-and-distance model asks me to learn average wait times, approximate route information, and consider potential worst case outcomes for both time and distance. It also requires me to police the route the cab driver is taking to ensure they aren't taking advantage of me. And I need to do all of this in every city I visit.
The upfront model asks me to be able to compare prices between different offerings (a skill I already have), and doesn't change at all between cities.
The whole Prop22 thing in California is another facet to the equation: ride sharing companies argue that freelance status means they have a far bigger, broader and elastic fleet, whereas a model with fixed costs would make availability look more like taxis before rideshare came along.
This isn't e-mail. The cost and hassle of switching to a competing service is nearly zero.
I mean, look at McDonalds and Coca-Cola, for example. I'm skeptical about the idea that simply having the SDV technology somehow also translates to being in a leadership position in terms of regulation, customer support, branding/marketing/being-installed-on-peoples-phones, etc etc.
Uber already had to fight uphill regarding regulation in many places so it definitely has some moat now (see, e.g. Uber and Ola in London). Waymo is on completely uncharted territory, regulation-wise.
I'm also curious about how the economics and logistics will play out. One criticism of existing ride sharing is not accounting for vehicle depreciation and wear and tear (and this on mass produced vehicles). Waymo vehicles have a lot more expensive technology on board, so I'm curious how that balances out in terms of lidar/etc shelf-life + specialized tech inspections vs ROI.
Also, cost of switching goes both ways. If this forces the hand of the other players to roll out SDV (via Cruise partnerships for example), Waymo doesn't necessarily stand to dominate the market, since it's already so crowded w/ well-funded/well-positioned players from all sorts of directions.
But at least for the next few years, technology is the barrier. So Google can partner with Lyft, for example, to get a good brand, app, customer service, regulation, etc.
But on top of that, the brand in this market is the promise "our cars are safe". Google is in a distinct place to own that, because of all the press they had, they press they will have, and by really leading technologically.
As for this being a commodity ? I believe we're really far from that.
I think a bigger challenge for Waymo is going to be getting trust and regulatory favor. Everyone keeps clinging on to the "oh, comparing prices, waymo is cheaper than uber because there's no driver!" narrative, but from what I've seen from the few reports so far, the prices aren't all that different from other services.
I can also imagine quite a few other not-so-flattering narratives being relatively realistic:
- "I'll never trust some faceless AI w/ my kids in the car! I'd rather pay a bit more and get a real person behind the wheel. Support the local economy, you know?"
- "Oh, so you want a license to put an entire class of workers out of jobs under MY government term?"
- "I got a trashed waymo car and customer support ignored me when I asked for a refund!"
- "I took a waymo after partying a bit too hard and now I'm locked out of all my google accounts with no recourse!"
- etc
I think people on HN often give way too much weight to the "cool tech" aspect of waymo, and disregard the very real and very significant meat bag aspect.
Current prices include a safety driver and small scale. Prices should go down in the future.
And sure, regulation and trust should be huge barriers. But besides the customer service issue(which seems too pessimistic), Google should be leading over others, on the trust issue(no accidents, more miles, highly skilled tech company), And on the regulation issue(good ties with Washington).
But those 2 are still a big risk, like you mention.
Not that particular company no, but the longevity of taxi service in general, yes. Although, it depends on why I'm riding in a taxi. If it's an occasional ride, it's not a big deal if taxi service disappears, I can pay for long term parking at the airport, or call a friend, or something. If it's an everyday ride, loss of service means a big change.
The danger of using a Google service is two-fold. The obvious part is, you might get used to it, only for it to be dropped. But also, they may run competitors out of business before dropping it, leaving you without any options.
Your cited risk is not realistic.
> The obvious part is, you might get used to it, only for it to be dropped.
> But also, they may run competitors out of business before dropping it, leaving you without any options.
It's a taxi. You call a number, or click a button on an app. A car arrives and drives you somewhere. The value delivered by the guy in the beat-up Prius, to me, is the exact same as Waymo. It's the most homogenized, commoditized industry in the world. Waymo could corner the entire market, and then go out of business, and the very next day, the streets would be full of independent taxis picking up passengers.
Some people care. They buy directly from the same families for many decades.
Serious now, no, Google does not get a break because they wish so. They have the right to shutter products and we have the right to bitch. Freedom at its best.