I'm convinced that Waymo will quietly continue to roll out self driving taxi's to ever increasing areas in "beta" until one day they're covering a significant percentage of the most profitable areas to be a taxi in.
Instead of trying for universal self driving, they're going to prioritise being self driving taxi's for the most profitable areas.
> 'm convinced that Waymo will quietly continue to roll out self driving taxi's to ever increasing areas in "beta" until one day they're covering a significant percentage of the most profitable areas to be a taxi in.
Makes sense. Ten cities are most of the US taxi market.[1] New York City alone is about a third.
New York will be tough in several directions, but Waymo will probably succeed there eventually.
Meanwhile, they can just scale up in LA.
I feel like some people have been using "most profitable areas" instead of saying dense urban areas and upset that it might take a while to reach rural and suburban areas. Is that what you are saying? What business is not in the business of pursuing the most plausible profitable direction first?
I think the general public has believed that self driving would arrive as some big bang moment where it just "worked" in 95%+ of environments where people drive. One morning there'd be an announcement it was cracked and then in six months you'd be able to buy a new car with it.
I don't think from either a business, technical or legal perspective google ever wants to sell or license their tech directly to a car being sold to the public.
They want to run a taxi service, they plan on running a taxi service, they're never going to sell it directly to consumers but they don't want consumers thinking about the implications of that.
They're just going to keep expanding Waymo in the background till it's ubiquitous.
Are they even profitable in SF or LA? I don't disagree with you, but if being profitable means charging more than Uber/Lyft and having slower trips on average than a human, thats not gonna work out too well. And these are the highest CoL cities in the us, meaning that if its not profitable there, might not be profitable anywhere.
Why do you think the trips are slower than a human? I've only taken one Waymo trip (and my wife two) but in each case it didn't feel any slower than a Lyft. But it felt a lot safer! They definitely drive better than any human Lyft driver I've ever had.
And the prices seems to roughly match Uber/Lyft. Sometimes they're higher sometimes lower. The one ride I took I chose the more expensive Waymo just for the better experience. You don't have to worry about what the driver might say or do that makes you uncomfortable or unsafe.
Their wait times tend to be worse, but that's getting better too. The trip speed has also noticeably improved. I've taken about 50 over the past 2 years.
Even if they're sometimes slower in pickup or trip time, on average, I greatly value the consistency of the experience over everything else.
Waymos are terrible at taking unprotected left turns into bumper to bumper traffic during peak hour.
They aren't aggressive like a normal driver when traffic is stopped and there's a clear but small gap that would result in the car being at an angle crossing traffic.
The lidar can't see any gap between cars when trying to turn into a more distant lane, so any obstruction to a close lane due to traffic kills them.
Source: been stuck in and observed Waymo in this situation for 10+ minutes several times.
A human driver is willing to make the turn much sooner.
Don't take a Waymo if you're in a hurry and the route will have such a left turns.
They seem to be getting more proactive in general. I was surprised last night when the one I was in went full barrel through a light that was already yellow and turned red while we were still in the intersection.
Ha, I had the exact same experience on a Waymo last night.
The light had just turned yellow before we entered the intersection, and I was sure it would slam the brakes (there was no one behind, so no risk of being read-ended). Yet it accelerated and cleared the intersection as the light turned red. It was what any reasonably good driver would do, but certainly edgy for a 100% law-abiding robot.
I’ll definitely take that over the Lyft driver I once had who went through a red light 1-2 seconds after it turned red. Not that that’s typical for ride share drivers in my area, but still.
Last Thursday. I called support after 10 minutes to cancel the ride but the car started moving while I was on the call.
That particular one was a traffic light green light with a "do not queue over intersection" area specifically designed to allow cars in my lane to turn onto the main road. The Waymo can't see that there's a gap in the lane it wanted to turn into, and was too conservative about queuing over the intersection at an angle when the light was green like any human would do.
I should first note that I'm a big fan of waymo and want autonomous to succeed generally.
I take both waymo's and lyft/uber all the time in sf and waymo's are way slower. I'd estimate it at 10-15% slower. Once the novelty of a waymo wears off you realize that they drive like a high anxiety teenager and going 15 mph on a 15 mph road, coming to gentle full stop at every stop sign, and being very tentative on turns and passing people all add up to a very slow ride.
The calmness is a major selling point for me, personally. Beyond just the safety aspect, I can't count how many times I've gotten nauseous from Uber and taxi drivers swerving and accelerating/stopping abruptly, often for no actual speed gain.
Cars of all kinds, really. But It seems to be more common in either crappy old gas cars or Teslas. I know Teslas have weird brake settings, but every ride I've experienced this has definitely been from aggressive and reckless driving.
Kidding but not kidding. Easily half my Uber rides have been legit scary. Yellow cabs only slightly better.
How does Waymo do for cleanliness and odor? That would be another selling point vs regular Uber/Lyft and many yellow cabs. These days they all seem to douse the interior with cheap cologne.
I stopped using Lyft after two close calls with drivers who must have faked their papers. One Lyft drive ran a red light into a postal truck at an otherwise empty intersection causing the front to fall off, like that Australian ship: https://www.youtube.com/watch?v=3m5qxZm_JqM
Stopping at a stop sign isn't just stopping at a stop sign. The way you pull up to it conveys a rough picture of your intentions to other drivers, or at least the ones that are on the ball. Good drivers actively time their arrival to avoid potential confusion with other traffic.
The sad truth about Waymo is that they have to aim for a 0 (or extremely close to 0) accident rate for PR/regulatory reasons, especially so for accidents caused by Waymo, and that's always going to heavily influence the way they drive.
Uber drivers cause accidents sometimes, just like all drivers do. Whenever this happens, we generally blame the driver and not Uber. It's a completely unremarkable event that the media will not pick up on, because everybody knows that car accidents are just a fact of life.
If a Waymo causes just one fatal accident, people will be up in arms and demand a ban on self driving. That means they have to be extremely conservative in how they drive, especially as more and more of them appear and the probability of one causing an accident goes up due to simple statistics.
There is a simpler explanation: every Waymo vehicle is effectively the same. If one makes a mistake, every vehicle will likely keep making the same mistake over and over again, until it's fixed. If the mistake causes harm, there is often a clear causal link from a flaw in the system to the harm, which can be a pretty good incentive to fix it.
Human drivers are all different. They also learn from their mistakes and change unpredictably over time. Humans can get away with all kinds of unsafe behavior, because bad outcomes are unlikely in any particular situation. If something bad happens, it could easily be a one-off issue. And even with systemic issues, it's easier to change the environment / regulations / vehicles than the drivers.
That's a product decision, though. They want people to feel safe, and to build a sustained record for being safe. They could easily hit the pedal if they wanted to.
"Lawful to a fault teenager" is not what people want in a taxi driver. That type of chauffeur doesn't make people feel safe. It makes them on edge. But unfortunately that's the kind of AI chauffeur regulators want.
I agree it's the "right" way to do it from a PR perspective though.
I'd rather do that than fear for my life when an uber driver decides that the 65mph speed limit should be understood as 85mph. Waymo doesn't even get on the freeway.
Operational profits is not really a goal for Waymo in near term. The current goal is to be many years ahead of competition, getting permits to operate in major profitable cities with profitable routes.
The cost of a Waymo car right now is around $150K. Soon, Waymo will build its own cars (with partnership with some car manufacturer) from scratch which will bring this cost to around $40K.
currently WayMo has around 2 people per car employed in service roles which will be like 2 people per 50 cars. At that point they will be swimming in profits.
> The current goal is to be many years ahead of competition
The problem with this is that one of their competitors (Tesla) now has very powerful friends in the government which are certainly not above using that power in their favor...
I'm pretty certain they won't undercut Uber or other taxi services any time soon because they don't want to provoke the political headache that comes with it. That type of politics could shutdown their licenses, I'm sure they want to avoid that at all costs so they're pricing above existing services.
What they may do is refuse to raise prices and eventually inflation will make them the cheapest.
They can charge more in Melbourne and I'll happily pay it to avoid drivers that cancel the trip after accepting or accept the trip and drive away from me until I cancel.
They're already the cheapest if they match prices because there is no tip required. In the long run they'll inevitably be cheaper because of higher utilization of their vehicles and (vastly) lower labor costs.
This is a misunderstanding of the economics of taxi firms. Taxi companies don't pay for purchase or depreciation of their cars (the drivers do), but Waymo does.
There are no drivers with Waymo. Believe it or not, a living annual wage for a person (or close to it) is vastly more expensive than the annual maintenance and depreciation on even a very expensive vehicle.
They're probably within reasonable distance of gross profit, and much farther from net profit because of the imbalance between their current fleet size and the massive ongoing manufacturing and R&D costs. The organizational goal is not immediate net profits though.
Take this with the grain of salt it deserves. No one relying solely on publicly available information can give you anything more than highly speculative guesses.
I honestly don't understand the appeal of a private self-driving vehicle as opposed to a taxi-style rental service.
I like that Waymo has human backup! Knowing that if the car gets into a tricky situation there's someone who can remote control it back out again is very reassuring.
I also like not having to think about where it parks, what it's doing when I'm not in it etc.
I would love to have one! Imagine all the benefits you just listed, but it's just mine so I can keep my stuff in it. Maybe my car seat or my work equipment or my shopping bags.
I get where I'm going and I send it over the the parking garage a mile away that's cheap, and then summon it back when I'm done with my event, so I never have to circle for parking. If enough people to it, they would probably develop a protocol so that they can be parked at a lot that is super dense and it's a puzzle to get the cars in and out.
And they could still be sold with remote human backup services (as well as giving you the owner the ability to take over if you want to).
I could also send it on errands. Go to Target and pick up my curbside order. Go to the pizza place and have them drop it on the front seat. Pick up my kid from school and bring her home. Go pick up my mother in law and bring her to my house so I don't have to go get her!
And I could use it to drive me to my parents house 350 miles away while I sleep.
In self driving world it no longer makes sense for a car to look like they do now.
Instead I think they operate a bit like Coach and Horses used to. You the consumer own the "coach", you have it specced out how you want it, your choice of seating (hell even a bed?), your sound system, your big screen tv, whatever you want because you're a passenger not the driver.
The "horses" are the self driving bit + batteries. The ubiquitous taxi service owns these. Maybe you permanently rent a low range model (100 miles) that is attached to your "Coach" at all times for quick errands. For longer range journeys you hire a model with more battery and longer range that drives over and attaches automatically to your coach.
For really long range journeys, your self driving "horse" drives into a coaching station along the way, detaches and another one attaches in 30 seconds and off you go.
The self driving bit has to be nailed down first I think, once that's done nothing I've described is even remotely difficult to implement, it's more a matter of peoples thinking around driving evolving.
I think it might go this way because people feel like they want to "own" their "cars". Especially in America it's a big part of the culture, so I can see being able to buy the coach part as being a popular compromise.
Powering electric cars is already cheaper, but right now they're less convenient, with self driving they become as convenient or more than gas vehicles. You no longer have to stop and charge anywhere because there's a fresh "horse" waiting to be hooked up while your existing one drives off to a charging station, all the downsides are automated away.
Maintaining an autonomous vehicle comes with a significantly higher maintenance burden than a regular car right now. Even if the sticker price were the same, would you be willing to check the air pressure regularly, take the vehicle in for calibration, clean the sensors, do software updates, implement the legal reporting responsibilities, etc? That's the current reality.
Most people don't want that and there's no market for it. Consumers (rightly) expect a vehicle that just works. It's still early days for this technology and building something that works as reliably/independently as the rest of a car isn't possible yet.
> Maintaining an autonomous vehicle comes with a significantly higher maintenance burden than a regular car right now.
Sure, and the cost would be really high too, but for a very wealthy person it might be worth it.
> would you be willing to check the air pressure regularly
I have to do that on my regular car, but it only needs air once every six months or so. New tires are pretty good about holding air. Also, no different than a regular car.
> take the vehicle in for calibration
I already have sensors on my car that require calibration, and I do that when it goes in for maintenance, about once a year. I wonder if these sensors need calibration more often.
> clean the sensors
You mean, wash the car? :)
> do software updates, implement the legal reporting responsibilities, etc?
I assume that would all be automatic over the air. And again, we're talking about someone who could afford a $150K car. There is a good chance they could hire someone to do that stuff too.
Generally, all of these (except sometimes calibration) are done on a weekly/monthly basis for testing fleets. Software updates and diagnostic reads can be automatic, but in practice there's enough moving parts that they require fairly regular intervention for safe operation. Legal reporting is not automatic currently. Tesla implemented that for their ADAS solutions and has been repeatedly grilled by NHTSA on their incomplete reporting. All of the commercial fleets implement reporting semi-manually because of the existential danger getting it wrong represents.
I think most people would rather have their car circle the block indefinitely, rather than pay for parking and wait for their car to summon, which would create so many new traffic problems.
I'm picturing herds of empty Model 3's endlessly circling every city's downtown office district
I've been poking at Uber forums and other places, and I'm noticing that a _lot_ of people are underestimating this. Waymo is growing by >100%/year at the moment and this is a significant part of that growth.
It'll make a huge dent in demand for Uber/Lyft over the next five years.
Unless those companies stumble on competing solutions that they can use to level the playing field. Waymo doesn't have this market to itself in perpetuity. Uber and Lyft basically have lots of revenue and can afford to take some bets on experimenting with different partners for self driving at a relatively low risk.
There are plenty of other companies experimenting with self driving taxis, cars, trucks, etc. And not just in the US. China is probably a lot further than Waymo on this front. In the end it boils down to cost at scale. Waymo has some early learnings for this, which is great, but it's no guarantee that somebody else won't come along and match what they are doing. Or massively improve on what and how they are doing what they are doing.
Tesla is the controversial one to mention in this context but I no longer believe that they'll get there first or exclusively. They might still get there. And maybe it won't even take that long. But they have a lot to prove and there are lots of other companies field testing working taxis with real passengers (unlike Tesla) at this point.
I wouldn't count Uber out. I'd say they are well aware of the challenges here and have a huge stake in ensuring that there will be multiple autonomy providers for them to choose from that all need the same thing to succeed: access to paying customers. Which is something that Uber has. The near future is likely to be mostly non-autonomous with specific geofences running experiments with autonomous driving in a limited number of places. Uber works in all those places and all they need is access to autonomous options in areas as that becomes available. They don't need to own or develop this in house. They just need access and some revenue share deal. More providers means more competitive deals.
I wouldn't be surprised to seem them close multiple, non-exclusive deals on this front in the next years. Possibly even with Waymo or Tesla. Or both. And then some.
Chinese autonomous deployments are pretty substantial at this point. And compared to places like Beijing, San Francisco is a bit of a small & rural village. China is rolling out at a faster pace than Waymo. More vehicles, more cities, more companies, larger cities, etc.
I wouldn't be surprised if Uber/Lyft essentially end up as aggregators for autonomous cars.
You'll have multiple (local) companies, possibly Uber franchisees, maintaining autonomous fleets and doing the low-margin work of car cleaning, acquiring the vehicles, servicing them etc. Uber will be the interface between the passenger and those companies.
Uber could even keep some human drivers for peak-traffic situations that it would be uneconomical for a ppurely autonomous fleet to handle.
I wonder if the waymo solution is very vehicle specific or if it could be easily applied to say, a 10~12 person van that could accommodate a wheelchair? It would be interesting to design a transit system around 4~5 times as many vehicles with higher interval or more responsive pathing with something like pick up kiosks.
Many cities are already upgrading their transit systems to implement "rapid ride" where they run buses on a 10min schedule through popular routes.
This eliminates the need for planning routes - go the nearest stop, wait 10-15min (sometimes buses get bunched up so it's not really every 10min), and get on the next bus.
Buses already have drivers, wheelchair (and bicycle!) support, etc. and the stops usually have live tracking of the next arriving buses.
I recently came across an article on the early years of the Zürich streetcar. In the early 1900s, 5 minute schedules were customary, with one especially busy line running a 3 minute schedule!
In contrast, nowadays the peak schedule is 7.5min.
Self-driving makes the most economic sense when fewer people are in a vehicle. With a large van or bus, the cost of the human driver is split between more passengers. Also van or bus service is less compelling to individuals, as it makes them beholden to bus stop locations and schedules instead of picking them up and dropping them off when and where they want.
Adding support for wheelchairs introduces new failure modes that would require more software and/or human intervention. For example: Wheelchairs need to be strapped down for safety, and not all wheelchair users can do this themselves.
> Also van or bus service is less compelling to individuals, as it makes them beholden to bus stop locations and schedules instead of picking them up and dropping them off when and where they want.
Exactly this. The attraction of self-driving transit is the support for individual planning. Everyone gets a direct route to their destination, no delays stopping or detouring for other people.
There's also increased safety in having a personal ride. You're not being hassled, stalked, or threatened by random people.
>That's an odd thing to say. The only times I've ever been randomly threatened in public were on the road, by another driver.
There's a reason we let drunks ride public transit but not drive.
The bar for conduct behind the wheel is higher than on the subway because the risks and stakes are much higher hence why people get pissed off much easier.
Not getting accosted on the subway is a simple task of not crossing paths with the few people doing the accosting. Not getting accosted on the road requires actively performing in a manner that keeps thing flowing smoothly.
It's the first non-US city that they've done public testing in. They've been doing test track work in Europe for awhile now, and Cruise did some Japanese/middle eastern testing before they folded.
> In addition to ongoing trips to Truckee, Michigan's Upper Peninsula, Upstate New York and Tokyo, the expansion includes testing in San Diego and Las Vegas, with more cities yet to be announced.
"In addition to ongoing trips to Truckee, Michigan's Upper Peninsula, Upstate New York and Tokyo, the expansion includes testing in San Diego and Las Vegas, with more cities yet to be announced."
Pitching my own book here, but I think Jackson Hole would be a great place for Waymo. It's clearly delineated, has basically no freeway driving, affluent residents and a constant influx of tourists for brand building. The only challenge is snow, but the roads are well plowed and drivers generally careful. If you're working in Truckee, I can't imagine it will be a much-bigger challenge.
I imagine they'll need a ton of testing and adjustments for driving in snow for all the conditions the car needs to detect and plan around, like the wheels slipping, or the car as a whole sliding down a hill. This is probably a separate effort from mapping out a city, but then needs to be tested in the real world.
I think Truckee and Tahoe offer this environment within reasonable driving distance to the main Waymo development campuses.
Not so much snow, as you have rapidly changing road conditions such as ice and freezing rain. And then factor in poor road markings and not always abundantly clear path finding.
The nice thing about EVs like they're using now. The electric motors are pretty good at responding and handling different road conditions, much faster than ICE vehicles since you can never quite predict what the engine is doing at a given moment.
Biggest difference is climate. Truckee's average highs never dip below 40F. Jackson has 2 months average highs below freezing -- that means a LOT more residual ice and snow.
theyd want to shut off on the odd snow day that makes lines invisible, but its a lot like phoenix on having big wide roads, and the places to consistently play with snow cover are pretty empty/safe to screw up in
> theyd want to shut off on the odd snow day that makes lines invisible
It'd sure suck if they go into cities run the local taxi companies and human drivers out of business only to shut down every time there's bad weather leaving people without access to transportation.
> they will push back because “NIMBYism” runs deep
Waymo is being adopted in wealthy areas first. To the extent there is NIMBY sentiment, it's towards rideshare drivers. (Also, billionaires probably wouldn't be taking Waymo. If their properties aren't fly in they're driving themselves.)
Yes, we agree. I meant in Jackson our billionaires either love driving, in which case they drive themselves, or don’t, in which case they fly straight in. The middle ground between those isn’t enough to found a market on.
Waymo has deep pockets and an incentive to play it safe. What it doesn't have is fierce competition forcing it to rush. Maybe that will change if/when Amazon's Zoox premiers this year.
I love that you included numbers, but that is for all of Alphabet's "other bets" which includes Waymo, so it's not entirely Waymo. It still helps give ballpark figures though.
Yeah, supposedly Verily had $559 MM in revenue in 2022, so total revenue for other bets (including Verily and Waymo) being 65% of that is pretty weird.
The subway already wins the competition with uber and drivers. 1 million drivers per day and 2 million riders per day. The subway is king . I love it so much.
I prefer streets that don’t have traffic so pedestrians can walk freely, and watching these waymo cars trap themselves and cause 30 minute long traffic jams because their code accidentally put them in a holding pattern is a genuine nightmare
If we want to exaggerate issues with each mode of transport, I prefer getting to my destination without getting stabbed, pushed onto the tracks, getting my nostrils assaulted by the smell of bodily fluids, having my ears assaulted by the train in the tunnels (14th/union sq 456 especially), having my ears assaulted by a showtimer, or waiting 20 mins for a train at 3am
Telling that Waymo is trying to steal the thunder by releasing this news on the same day as Tesla's Q4 earnings report. They see the competition about to overtake them.
I hate ADS tech in cities so much. The pedestrian/pet detection is god awful. Doesn’t matter if you cross at designated crosswalks or just crossing the street. Many close calls.
Cruise _was_ the king in this aspect and avoided them completely but Waymo is getting up there.
Side note: GOOG and other tech conglomerates are clearly monopolizing the tech industry - from ADS to AI/AGI/LLM. It’s no wonder these companies have kissed the ring of current administration to get the green light on acquisitions needed to setup monopolies and effectively become a monopsony (particularly in labor to keep wages down).
Would any smaller of a company had the funds to acquihire an entire darpa challenge team and have them seemingly fruitlessly work on autonomous driving for over 10 years? We have already seen what happens when the parent company is any less patient and deep pocketed, with cruise and GM.
When I walk in front of a Waymo it stops and gives me space and doesn't come anywhere near hitting me. That doesn't qualify as "god awful" to me. do you have a different experience?
I'm convinced that Waymo will quietly continue to roll out self driving taxi's to ever increasing areas in "beta" until one day they're covering a significant percentage of the most profitable areas to be a taxi in.
Instead of trying for universal self driving, they're going to prioritise being self driving taxi's for the most profitable areas.
> 'm convinced that Waymo will quietly continue to roll out self driving taxi's to ever increasing areas in "beta" until one day they're covering a significant percentage of the most profitable areas to be a taxi in.
Makes sense. Ten cities are most of the US taxi market.[1] New York City alone is about a third. New York will be tough in several directions, but Waymo will probably succeed there eventually. Meanwhile, they can just scale up in LA.
[1] https://en.wikipedia.org/wiki/Taxis_of_the_United_States
I feel like some people have been using "most profitable areas" instead of saying dense urban areas and upset that it might take a while to reach rural and suburban areas. Is that what you are saying? What business is not in the business of pursuing the most plausible profitable direction first?
I think the general public has believed that self driving would arrive as some big bang moment where it just "worked" in 95%+ of environments where people drive. One morning there'd be an announcement it was cracked and then in six months you'd be able to buy a new car with it.
I don't think from either a business, technical or legal perspective google ever wants to sell or license their tech directly to a car being sold to the public.
They want to run a taxi service, they plan on running a taxi service, they're never going to sell it directly to consumers but they don't want consumers thinking about the implications of that.
They're just going to keep expanding Waymo in the background till it's ubiquitous.
Most startups actually.
Are they even profitable in SF or LA? I don't disagree with you, but if being profitable means charging more than Uber/Lyft and having slower trips on average than a human, thats not gonna work out too well. And these are the highest CoL cities in the us, meaning that if its not profitable there, might not be profitable anywhere.
Why do you think the trips are slower than a human? I've only taken one Waymo trip (and my wife two) but in each case it didn't feel any slower than a Lyft. But it felt a lot safer! They definitely drive better than any human Lyft driver I've ever had.
And the prices seems to roughly match Uber/Lyft. Sometimes they're higher sometimes lower. The one ride I took I chose the more expensive Waymo just for the better experience. You don't have to worry about what the driver might say or do that makes you uncomfortable or unsafe.
Their wait times tend to be worse, but that's getting better too. The trip speed has also noticeably improved. I've taken about 50 over the past 2 years.
Even if they're sometimes slower in pickup or trip time, on average, I greatly value the consistency of the experience over everything else.
Waymos are terrible at taking unprotected left turns into bumper to bumper traffic during peak hour.
They aren't aggressive like a normal driver when traffic is stopped and there's a clear but small gap that would result in the car being at an angle crossing traffic.
The lidar can't see any gap between cars when trying to turn into a more distant lane, so any obstruction to a close lane due to traffic kills them.
Source: been stuck in and observed Waymo in this situation for 10+ minutes several times.
A human driver is willing to make the turn much sooner.
Don't take a Waymo if you're in a hurry and the route will have such a left turns.
Anecdotally, they've gotten much more proactive at taking unprotected lefts in my experience.
They seem to be getting more proactive in general. I was surprised last night when the one I was in went full barrel through a light that was already yellow and turned red while we were still in the intersection.
Ha, I had the exact same experience on a Waymo last night.
The light had just turned yellow before we entered the intersection, and I was sure it would slam the brakes (there was no one behind, so no risk of being read-ended). Yet it accelerated and cleared the intersection as the light turned red. It was what any reasonably good driver would do, but certainly edgy for a 100% law-abiding robot.
Pretty flawless experience.
I’ll definitely take that over the Lyft driver I once had who went through a red light 1-2 seconds after it turned red. Not that that’s typical for ride share drivers in my area, but still.
When was the last time this happened to you?
Last Thursday. I called support after 10 minutes to cancel the ride but the car started moving while I was on the call.
That particular one was a traffic light green light with a "do not queue over intersection" area specifically designed to allow cars in my lane to turn onto the main road. The Waymo can't see that there's a gap in the lane it wanted to turn into, and was too conservative about queuing over the intersection at an angle when the light was green like any human would do.
I should first note that I'm a big fan of waymo and want autonomous to succeed generally.
I take both waymo's and lyft/uber all the time in sf and waymo's are way slower. I'd estimate it at 10-15% slower. Once the novelty of a waymo wears off you realize that they drive like a high anxiety teenager and going 15 mph on a 15 mph road, coming to gentle full stop at every stop sign, and being very tentative on turns and passing people all add up to a very slow ride.
You're right though it definitely feels safer.
The calmness is a major selling point for me, personally. Beyond just the safety aspect, I can't count how many times I've gotten nauseous from Uber and taxi drivers swerving and accelerating/stopping abruptly, often for no actual speed gain.
Were those rides in hybrids? I've noticed the way they brake is a lot more likely to get me carsick if I'm sitting in the back, especially Priuses.
Cars of all kinds, really. But It seems to be more common in either crappy old gas cars or Teslas. I know Teslas have weird brake settings, but every ride I've experienced this has definitely been from aggressive and reckless driving.
It stopped at a stop sign?!?!? THE HORROR!
Kidding but not kidding. Easily half my Uber rides have been legit scary. Yellow cabs only slightly better.
How does Waymo do for cleanliness and odor? That would be another selling point vs regular Uber/Lyft and many yellow cabs. These days they all seem to douse the interior with cheap cologne.
I stopped using Lyft after two close calls with drivers who must have faked their papers. One Lyft drive ran a red light into a postal truck at an otherwise empty intersection causing the front to fall off, like that Australian ship: https://www.youtube.com/watch?v=3m5qxZm_JqM
Stopping at a stop sign isn't just stopping at a stop sign. The way you pull up to it conveys a rough picture of your intentions to other drivers, or at least the ones that are on the ball. Good drivers actively time their arrival to avoid potential confusion with other traffic.
The sad truth about Waymo is that they have to aim for a 0 (or extremely close to 0) accident rate for PR/regulatory reasons, especially so for accidents caused by Waymo, and that's always going to heavily influence the way they drive.
Uber drivers cause accidents sometimes, just like all drivers do. Whenever this happens, we generally blame the driver and not Uber. It's a completely unremarkable event that the media will not pick up on, because everybody knows that car accidents are just a fact of life.
If a Waymo causes just one fatal accident, people will be up in arms and demand a ban on self driving. That means they have to be extremely conservative in how they drive, especially as more and more of them appear and the probability of one causing an accident goes up due to simple statistics.
There is a simpler explanation: every Waymo vehicle is effectively the same. If one makes a mistake, every vehicle will likely keep making the same mistake over and over again, until it's fixed. If the mistake causes harm, there is often a clear causal link from a flaw in the system to the harm, which can be a pretty good incentive to fix it.
Human drivers are all different. They also learn from their mistakes and change unpredictably over time. Humans can get away with all kinds of unsafe behavior, because bad outcomes are unlikely in any particular situation. If something bad happens, it could easily be a one-off issue. And even with systemic issues, it's easier to change the environment / regulations / vehicles than the drivers.
That's a product decision, though. They want people to feel safe, and to build a sustained record for being safe. They could easily hit the pedal if they wanted to.
They are doing the right thing.
>They want people to feel safe.
"Lawful to a fault teenager" is not what people want in a taxi driver. That type of chauffeur doesn't make people feel safe. It makes them on edge. But unfortunately that's the kind of AI chauffeur regulators want.
I agree it's the "right" way to do it from a PR perspective though.
They don't drive like teenagers do (the Tesla FSD does, but that's another story).
A Waymo drives more like a 50 year old who's seen some shit and drives well but defensively.
I'd rather do that than fear for my life when an uber driver decides that the 65mph speed limit should be understood as 85mph. Waymo doesn't even get on the freeway.
Operational profits is not really a goal for Waymo in near term. The current goal is to be many years ahead of competition, getting permits to operate in major profitable cities with profitable routes.
The cost of a Waymo car right now is around $150K. Soon, Waymo will build its own cars (with partnership with some car manufacturer) from scratch which will bring this cost to around $40K.
currently WayMo has around 2 people per car employed in service roles which will be like 2 people per 50 cars. At that point they will be swimming in profits.
> The current goal is to be many years ahead of competition
The problem with this is that one of their competitors (Tesla) now has very powerful friends in the government which are certainly not above using that power in their favor...
This is another reason why I think they will remain low key.
They'll perpetually be in "beta" until they're so ubiquitous that the idea of getting rid of them is unthinkable.
> currently WayMo has around 2 people per car employed in service roles
Do you have a source for this? I'm curious about their operations
It's almost certainly not true.
I'm pretty certain they won't undercut Uber or other taxi services any time soon because they don't want to provoke the political headache that comes with it. That type of politics could shutdown their licenses, I'm sure they want to avoid that at all costs so they're pricing above existing services.
What they may do is refuse to raise prices and eventually inflation will make them the cheapest.
They can charge more in Melbourne and I'll happily pay it to avoid drivers that cancel the trip after accepting or accept the trip and drive away from me until I cancel.
Charging about as much as their competitors is probably the profit-maximizing strategy.
They're already the cheapest if they match prices because there is no tip required. In the long run they'll inevitably be cheaper because of higher utilization of their vehicles and (vastly) lower labor costs.
This is a misunderstanding of the economics of taxi firms. Taxi companies don't pay for purchase or depreciation of their cars (the drivers do), but Waymo does.
There are no drivers with Waymo. Believe it or not, a living annual wage for a person (or close to it) is vastly more expensive than the annual maintenance and depreciation on even a very expensive vehicle.
They're probably within reasonable distance of gross profit, and much farther from net profit because of the imbalance between their current fleet size and the massive ongoing manufacturing and R&D costs. The organizational goal is not immediate net profits though.
Take this with the grain of salt it deserves. No one relying solely on publicly available information can give you anything more than highly speculative guesses.
Maybe not profitable yet. But each generation Waymo car seems to cost about half of the previous generation.
I would pay more for a slower ride just to avoid having to pay a tip.
I will pay more to avoid having to talk to the driver. Last taxi I took, the driver started talking to me about all the different world religions.
You will certainly be happy to learn that Waymo is in fact introducing tipping.
https://x.com/wongmjane/status/1884136644238008334
It goes to a charity of the rider's choice [1].
[1] https://techcrunch.com/2025/01/28/waymo-is-reportedly-buildi...
>but if being profitable means charging more than Uber/Lyft and having slower trips on average than a human
thats just a function of the numbers of cars on the road
I honestly don't understand the appeal of a private self-driving vehicle as opposed to a taxi-style rental service.
I like that Waymo has human backup! Knowing that if the car gets into a tricky situation there's someone who can remote control it back out again is very reassuring.
I also like not having to think about where it parks, what it's doing when I'm not in it etc.
I would love to have one! Imagine all the benefits you just listed, but it's just mine so I can keep my stuff in it. Maybe my car seat or my work equipment or my shopping bags.
I get where I'm going and I send it over the the parking garage a mile away that's cheap, and then summon it back when I'm done with my event, so I never have to circle for parking. If enough people to it, they would probably develop a protocol so that they can be parked at a lot that is super dense and it's a puzzle to get the cars in and out.
And they could still be sold with remote human backup services (as well as giving you the owner the ability to take over if you want to).
I could also send it on errands. Go to Target and pick up my curbside order. Go to the pizza place and have them drop it on the front seat. Pick up my kid from school and bring her home. Go pick up my mother in law and bring her to my house so I don't have to go get her!
And I could use it to drive me to my parents house 350 miles away while I sleep.
I can think of a million reasons I'd want one.
I've been thinking about this.
In self driving world it no longer makes sense for a car to look like they do now.
Instead I think they operate a bit like Coach and Horses used to. You the consumer own the "coach", you have it specced out how you want it, your choice of seating (hell even a bed?), your sound system, your big screen tv, whatever you want because you're a passenger not the driver.
The "horses" are the self driving bit + batteries. The ubiquitous taxi service owns these. Maybe you permanently rent a low range model (100 miles) that is attached to your "Coach" at all times for quick errands. For longer range journeys you hire a model with more battery and longer range that drives over and attaches automatically to your coach.
For really long range journeys, your self driving "horse" drives into a coaching station along the way, detaches and another one attaches in 30 seconds and off you go.
Ooh interesting vision! Are any companies pursuing this avenue?
The self driving bit has to be nailed down first I think, once that's done nothing I've described is even remotely difficult to implement, it's more a matter of peoples thinking around driving evolving.
I think it might go this way because people feel like they want to "own" their "cars". Especially in America it's a big part of the culture, so I can see being able to buy the coach part as being a popular compromise.
Powering electric cars is already cheaper, but right now they're less convenient, with self driving they become as convenient or more than gas vehicles. You no longer have to stop and charge anywhere because there's a fresh "horse" waiting to be hooked up while your existing one drives off to a charging station, all the downsides are automated away.
Maintaining an autonomous vehicle comes with a significantly higher maintenance burden than a regular car right now. Even if the sticker price were the same, would you be willing to check the air pressure regularly, take the vehicle in for calibration, clean the sensors, do software updates, implement the legal reporting responsibilities, etc? That's the current reality.
Most people don't want that and there's no market for it. Consumers (rightly) expect a vehicle that just works. It's still early days for this technology and building something that works as reliably/independently as the rest of a car isn't possible yet.
> Maintaining an autonomous vehicle comes with a significantly higher maintenance burden than a regular car right now.
Sure, and the cost would be really high too, but for a very wealthy person it might be worth it.
> would you be willing to check the air pressure regularly
I have to do that on my regular car, but it only needs air once every six months or so. New tires are pretty good about holding air. Also, no different than a regular car.
> take the vehicle in for calibration
I already have sensors on my car that require calibration, and I do that when it goes in for maintenance, about once a year. I wonder if these sensors need calibration more often.
> clean the sensors
You mean, wash the car? :)
> do software updates, implement the legal reporting responsibilities, etc?
I assume that would all be automatic over the air. And again, we're talking about someone who could afford a $150K car. There is a good chance they could hire someone to do that stuff too.
Generally, all of these (except sometimes calibration) are done on a weekly/monthly basis for testing fleets. Software updates and diagnostic reads can be automatic, but in practice there's enough moving parts that they require fairly regular intervention for safe operation. Legal reporting is not automatic currently. Tesla implemented that for their ADAS solutions and has been repeatedly grilled by NHTSA on their incomplete reporting. All of the commercial fleets implement reporting semi-manually because of the existential danger getting it wrong represents.
I think most people would rather have their car circle the block indefinitely, rather than pay for parking and wait for their car to summon, which would create so many new traffic problems.
I'm picturing herds of empty Model 3's endlessly circling every city's downtown office district
> but it's just mine so I can keep my stuff in it
In the distant future (10 years?) we may have private trailers/pods that are moved by self-driving vehicles. Best of both options.
They're already covering 0.25% of total US paved road network per my napkin calculation.
yeah, that's exactly what they are going to do. It's how Uber worked as well.
I've been poking at Uber forums and other places, and I'm noticing that a _lot_ of people are underestimating this. Waymo is growing by >100%/year at the moment and this is a significant part of that growth.
It'll make a huge dent in demand for Uber/Lyft over the next five years.
Unless those companies stumble on competing solutions that they can use to level the playing field. Waymo doesn't have this market to itself in perpetuity. Uber and Lyft basically have lots of revenue and can afford to take some bets on experimenting with different partners for self driving at a relatively low risk.
There are plenty of other companies experimenting with self driving taxis, cars, trucks, etc. And not just in the US. China is probably a lot further than Waymo on this front. In the end it boils down to cost at scale. Waymo has some early learnings for this, which is great, but it's no guarantee that somebody else won't come along and match what they are doing. Or massively improve on what and how they are doing what they are doing.
Tesla is the controversial one to mention in this context but I no longer believe that they'll get there first or exclusively. They might still get there. And maybe it won't even take that long. But they have a lot to prove and there are lots of other companies field testing working taxis with real passengers (unlike Tesla) at this point.
I wouldn't count Uber out. I'd say they are well aware of the challenges here and have a huge stake in ensuring that there will be multiple autonomy providers for them to choose from that all need the same thing to succeed: access to paying customers. Which is something that Uber has. The near future is likely to be mostly non-autonomous with specific geofences running experiments with autonomous driving in a limited number of places. Uber works in all those places and all they need is access to autonomous options in areas as that becomes available. They don't need to own or develop this in house. They just need access and some revenue share deal. More providers means more competitive deals.
I wouldn't be surprised to seem them close multiple, non-exclusive deals on this front in the next years. Possibly even with Waymo or Tesla. Or both. And then some.
> Uber and Lyft basically have lots of revenue
Waymo is level with Lyft in San Francisco [1].
> plenty of other companies experimenting with self driving taxis, cars, trucks, etc.
None with a track record. The only wild card is Tesla.
[1] https://www.ktvu.com/news/waymo-ridership-skyrocketed-in-sf-...
Chinese autonomous deployments are pretty substantial at this point. And compared to places like Beijing, San Francisco is a bit of a small & rural village. China is rolling out at a faster pace than Waymo. More vehicles, more cities, more companies, larger cities, etc.
I wouldn't be surprised if Uber/Lyft essentially end up as aggregators for autonomous cars.
You'll have multiple (local) companies, possibly Uber franchisees, maintaining autonomous fleets and doing the low-margin work of car cleaning, acquiring the vehicles, servicing them etc. Uber will be the interface between the passenger and those companies.
Uber could even keep some human drivers for peak-traffic situations that it would be uneconomical for a ppurely autonomous fleet to handle.
Interestingly, they're currently partnering with Uber in Florida if I remember correctly.
Source: https://www.theverge.com/news/600542/waymo-test-cities-las-v...
I wonder if the waymo solution is very vehicle specific or if it could be easily applied to say, a 10~12 person van that could accommodate a wheelchair? It would be interesting to design a transit system around 4~5 times as many vehicles with higher interval or more responsive pathing with something like pick up kiosks.
Many cities are already upgrading their transit systems to implement "rapid ride" where they run buses on a 10min schedule through popular routes.
This eliminates the need for planning routes - go the nearest stop, wait 10-15min (sometimes buses get bunched up so it's not really every 10min), and get on the next bus.
Buses already have drivers, wheelchair (and bicycle!) support, etc. and the stops usually have live tracking of the next arriving buses.
I recently came across an article on the early years of the Zürich streetcar. In the early 1900s, 5 minute schedules were customary, with one especially busy line running a 3 minute schedule!
In contrast, nowadays the peak schedule is 7.5min.
Self-driving makes the most economic sense when fewer people are in a vehicle. With a large van or bus, the cost of the human driver is split between more passengers. Also van or bus service is less compelling to individuals, as it makes them beholden to bus stop locations and schedules instead of picking them up and dropping them off when and where they want.
Adding support for wheelchairs introduces new failure modes that would require more software and/or human intervention. For example: Wheelchairs need to be strapped down for safety, and not all wheelchair users can do this themselves.
> Also van or bus service is less compelling to individuals, as it makes them beholden to bus stop locations and schedules instead of picking them up and dropping them off when and where they want.
Exactly this. The attraction of self-driving transit is the support for individual planning. Everyone gets a direct route to their destination, no delays stopping or detouring for other people.
There's also increased safety in having a personal ride. You're not being hassled, stalked, or threatened by random people.
> You're not being hassled, stalked, or threatened by random people.
That's an odd thing to say. The only times I've ever been randomly threatened in public were on the road, by another driver.
Your privilege is showing ;) A quick search easily found some articles for learning more about the topic:
“You’re just constantly on alert”: Women and Gender-Diverse People’s Experiences of Sexual Violence on Public Transport -> https://pmc.ncbi.nlm.nih.gov/articles/PMC10515455/
Uncovering gender-based violence and harassment in public transport: Lessons for spatial and transport justice -> https://www.sciencedirect.com/science/article/abs/pii/S09666...
>That's an odd thing to say. The only times I've ever been randomly threatened in public were on the road, by another driver.
There's a reason we let drunks ride public transit but not drive.
The bar for conduct behind the wheel is higher than on the subway because the risks and stakes are much higher hence why people get pissed off much easier.
Not getting accosted on the subway is a simple task of not crossing paths with the few people doing the accosting. Not getting accosted on the road requires actively performing in a manner that keeps thing flowing smoothly.
> With a large van or bus, the cost of the human driver is split between more passengers
It's also increased in that one must manage the passengers.
I'm seeing a lot of talk of there being a shortage of bus drivers in many places, so self-driving could probably still help increase service
That’s offset by dramatically lower prices and significant service in off peak hours.
Net result bus drivers are still a large fraction of the cost of bus service.
Are you... is this the micro-bus?
Mass manufacturing vehicles is important part of their long term strategy. So I am pretty sure a large van with wheelchair support will be on cards.
Tokyo is one of the new cities, interesting. Is this the first non-US city waymo has tested in?
It's the first non-US city that they've done public testing in. They've been doing test track work in Europe for awhile now, and Cruise did some Japanese/middle eastern testing before they folded.
Thats a great choice. High trust city, super high density.
Well, low crime, not high trust. Like, noone trusts each other to throw out their trash the right way.
The article does not actually list the 10 cities. Nor does it link to a source or say the source.
> In addition to ongoing trips to Truckee, Michigan's Upper Peninsula, Upstate New York and Tokyo, the expansion includes testing in San Diego and Las Vegas, with more cities yet to be announced.
They list a few cities but not 10 cities.
Presumably this is counting Atlanta, Miami, and Austin - which were already announced as expansion markets for 2025.
See "coming soon" on the WayMo homepage: https://waymo.com/
That's 9 total.
> with more cities yet to be announced
Anyone know how it does in wintery places? I'm seeing that it's in Michigan now
"In addition to ongoing trips to Truckee, Michigan's Upper Peninsula, Upstate New York and Tokyo, the expansion includes testing in San Diego and Las Vegas, with more cities yet to be announced."
Pitching my own book here, but I think Jackson Hole would be a great place for Waymo. It's clearly delineated, has basically no freeway driving, affluent residents and a constant influx of tourists for brand building. The only challenge is snow, but the roads are well plowed and drivers generally careful. If you're working in Truckee, I can't imagine it will be a much-bigger challenge.
I imagine they'll need a ton of testing and adjustments for driving in snow for all the conditions the car needs to detect and plan around, like the wheels slipping, or the car as a whole sliding down a hill. This is probably a separate effort from mapping out a city, but then needs to be tested in the real world.
I think Truckee and Tahoe offer this environment within reasonable driving distance to the main Waymo development campuses.
They're testing in Upper Minnesota and upstate New York. Upstate New York contains the snowiest city in the US, Syracuse.
Ah, good point then!
Not so much snow, as you have rapidly changing road conditions such as ice and freezing rain. And then factor in poor road markings and not always abundantly clear path finding.
The nice thing about EVs like they're using now. The electric motors are pretty good at responding and handling different road conditions, much faster than ICE vehicles since you can never quite predict what the engine is doing at a given moment.
Truckee/Tahoe is an extremely popular test area for bay area AV companies. Waymo's been testing there regularly since at least 2017.
Biggest difference is climate. Truckee's average highs never dip below 40F. Jackson has 2 months average highs below freezing -- that means a LOT more residual ice and snow.
calgary could be a good spot
theyd want to shut off on the odd snow day that makes lines invisible, but its a lot like phoenix on having big wide roads, and the places to consistently play with snow cover are pretty empty/safe to screw up in
> theyd want to shut off on the odd snow day that makes lines invisible
It'd sure suck if they go into cities run the local taxi companies and human drivers out of business only to shut down every time there's bad weather leaving people without access to transportation.
I didn’t realize there is a private equity and investing scene in Jackson Hole.
> think Jackson Hole would be a great place for Waymo
billionaire class wouldn’t be too happy with getting beta tested on raw tech. They will push back because “NIMBYism” runs deep.
> they will push back because “NIMBYism” runs deep
Waymo is being adopted in wealthy areas first. To the extent there is NIMBY sentiment, it's towards rideshare drivers. (Also, billionaires probably wouldn't be taking Waymo. If their properties aren't fly in they're driving themselves.)
In SF, a lot of billionaires love taking Waymos
Yes, we agree. I meant in Jackson our billionaires either love driving, in which case they drive themselves, or don’t, in which case they fly straight in. The middle ground between those isn’t enough to found a market on.
When will this be rolled out to everywhere else?
Waymo has deep pockets and an incentive to play it safe. What it doesn't have is fierce competition forcing it to rush. Maybe that will change if/when Amazon's Zoox premiers this year.
In 2024 it had $365 million in revenue but $1.13 billion expenses.
I love that you included numbers, but that is for all of Alphabet's "other bets" which includes Waymo, so it's not entirely Waymo. It still helps give ballpark figures though.
isn't waymo the majority of alphabets other bets?
In terms of revenue - but less so in terms of costs.
I wouldn't be surprised if WayMo is losing <$200M instead of ~$800M like the above post made it seem.
Additionally - the majority of those losses could come from expansion efforts.
I wouldn't be surprised if WayMo already has profitable unit economics.
verily
Yeah, supposedly Verily had $559 MM in revenue in 2022, so total revenue for other bets (including Verily and Waymo) being 65% of that is pretty weird.
August 9th 2038. 11:42am.
Zulu time.
Good - a few months after Y2K38 so they'll have worked out all the problems by then.
You bring that crap to nyc and we’re making a special new waymo congestion price
I can’t wait until this hits NYC and the subway is forced to compete on all 3 of price, comfort, convenience
The subway already wins the competition with uber and drivers. 1 million drivers per day and 2 million riders per day. The subway is king . I love it so much.
Pull up the MTA tracks and let Waymo better use the right of way.
So you prefer human taxi drivers?
I prefer streets that don’t have traffic so pedestrians can walk freely, and watching these waymo cars trap themselves and cause 30 minute long traffic jams because their code accidentally put them in a holding pattern is a genuine nightmare
If we want to exaggerate issues with each mode of transport, I prefer getting to my destination without getting stabbed, pushed onto the tracks, getting my nostrils assaulted by the smell of bodily fluids, having my ears assaulted by the train in the tunnels (14th/union sq 456 especially), having my ears assaulted by a showtimer, or waiting 20 mins for a train at 3am
Right. This will never be improved
Telling that Waymo is trying to steal the thunder by releasing this news on the same day as Tesla's Q4 earnings report. They see the competition about to overtake them.
I hate ADS tech in cities so much. The pedestrian/pet detection is god awful. Doesn’t matter if you cross at designated crosswalks or just crossing the street. Many close calls.
Cruise _was_ the king in this aspect and avoided them completely but Waymo is getting up there.
Side note: GOOG and other tech conglomerates are clearly monopolizing the tech industry - from ADS to AI/AGI/LLM. It’s no wonder these companies have kissed the ring of current administration to get the green light on acquisitions needed to setup monopolies and effectively become a monopsony (particularly in labor to keep wages down).
Would any smaller of a company had the funds to acquihire an entire darpa challenge team and have them seemingly fruitlessly work on autonomous driving for over 10 years? We have already seen what happens when the parent company is any less patient and deep pocketed, with cruise and GM.
> The pedestrian/pet detection is god awful
When I walk in front of a Waymo it stops and gives me space and doesn't come anywhere near hitting me. That doesn't qualify as "god awful" to me. do you have a different experience?