Posted by doctoboggan 3 days ago
Of course, people won't like this, I'm not exactly enthused either, but the alternative would be a corporation constantly providing -- for free -- updates and even support if your car gets into an accident or stuck. That doesn't really make sense from a business perspective.
1. Courts are finding Tesla partially liable for collisions, so they've already got some of the downsides of insurance (aka the payout) without the upside (the premium).
2. Waymo data shows a significant injury reduction rate. If it's true and not manipulated data, it's natural for the car companies to want to capture some of this upside.
3. It just seems like a much easier sell. I wouldn't pay $100/month for self-driving, but $150 a month for self-driving + insurance? That's more than I currently pay for insurance, but not a lot more. And I've got relatively cheap insurance: charging $250/month for insurance + self-driving will be cheaper than what some people pay for just insurance alone.
I don't think we need to hit 100% self-driving for the bundled insurance to be viable. 90% self-driving should still have a substantially lower accident rate if the Waymo data is accurate and extends.
1. High-severity accidents might drop, but the industry bleeds money on high-frequency, low-speed incidents (parking lots, neighborhood scrapes). Autonomy has diminishing returns here; it doesn't magically prevent the chaos of mixed-use environments.
2. Insurance is a capital management game. We’ll likely see a tech company try this, fail to cover a catastrophic liability due to lack of reserves, and trigger a massive backlash.
It reminds me of early internet optimism: we thought connectivity would make truth impossible to hide. Instead, we got the opposite. Tech rarely solves complex markets linearly.
Google, AFAIK the only company with cars that are actually autonomous, has US$98 Billion in cash.
It'd have to be a hell of an accident to put a dent in that.
My guess, if this actually plays out, is that existing insurers will create a special autonomy product that will modify rates to reflect differences in risk from standard driving, and autonomy subscriptions will offer those in a bundle.
It doesn't prevent chaos, but it does provide ubiquitous cameras. That will be used against people.
I'm ambivalent about that and mostly in a negative direction. On the one hand, I'd very much love to see people who cause accidents have their insurance go through the roof.
On the other hand, the insurance companies will force self-driving on everybody through massive insurance rate increases for manual driving. Given that we do not have protections against companies that can make you a Digital Non-Person with a click of a mouse, I have significant problems with that.
Life is hard and people make mistakes. Let the actuaries do their job, but causing an accident is not a moral failure, except in cases like drunk driving, where we have actual criminal liability already.
> the insurance companies will force self-driving on everybody through massive insurance rate increases for manual driving.
Why would manual driving be more expensive to insure in the future? The same risks exist today, at today's rates, but with the benefit that over time the other cars will get harder to hit, reducing the rate of accidents even for humans (kinda like herd immunity).
> Given that we do not have protections against companies that can make you a Digital Non-Person with a click of a mouse, I have significant problems with that.
I absolutely think this is going to be one of the greater social issues of the next generation.
This seems like it can be solved with a deductible.
First, if the insurance applies to fully autonomous driving only, then I suspect they’ll reach a point where the cost of insurance+automation ends up being less than just insurance through third parties.
Second, cutting into the traditional insurance market share is likely to increase costs for those who remain on traditional insurance, assuming there’s a significant enough number of people jumping ship. Combined, this creates a huge incentive for more users to jump on the self-driving bandwagon.
Keyword: my own SUV. Not a rental. With the possibility for me to take over and drive it myself if service fails or if I want to do so.
The significant unlock is that I get to haul gear, packages, family. I don't need to keep it clean. The muddy dogs, the hiking trip, the week-long road trip.
If my car could drive me, I'd do way more road trips and skip flying. It's almost as romantic as a California Zephyr or Coast Starlight trip. And I can camp out of it.
No cramped airlines. No catching colds by being packed in a sardine can with a stressed out immune system.
No sharing space with people on public transit. I can work and watch movies and listen to music and hang out with my wife, my friends. People won't stare at me, and I can eat in peace or just be myself in my own space.
I might even work in a nomadic lifestyle if I don't have to drive all the time. Our country is so big and there's so much to see.
One day you might even be able to attach a trailer. Bikes, jet skis, ATVs. People might simply live on the road, traveling all the time.
Big cars seem preferable. Lots of space for internal creature comforts. Laying back, lounging. Watching, reading, eating. Changing clothes, camping, even cooking.
Some people might even buy autonomous RVs. I'm sure that'll be a big thing in its own right.
It's bidirectional too! People can come to you as you go to them. Meet in the middle. Same thing with packages, food, etc.
This would be the biggest thing in travel, transport, logistics, perhaps ever. It's a huge unlock. It feels downright revolutionary. Like a total change in how we might live our lives.
This might turn big suburbs from food/culture deserts into the default places people want to live as they have more space for cheaper - because the commute falls apart.
This honestly sounds better than a house, but if you can also own an affordable large home in the suburbs as your home base - that's incredible. You don't need a tiny expensive place in the city. You could fall asleep in your car and wake up for breakfast in the city. Spend some time at home, then make a trek to the mountains. All without wasting any time. No more driving, no more traffic. Commuting becomes leisure. It becomes you time.
This is also kind of a super power that big countries (in terms of area) with lots of roads and highways will enjoy the most. It doesn't do much in a dense city, but once you add mountains and forests and streams and deserts and oceans - that's magic.
Maybe our vast interstate highway infrastructure will suddenly grow ten times in value.
Roads might become more important than ever. We might even start building more.
If the insurance and autonomy come bundled as a subscription after you purchase or lease your vehicle, that's super easy for people to activate and spend money on.
This is such a romantic dream, and I'm so hyped for this.
I would pay an ungodly sum to unlock this. It can't come soon enough. Would subscribe in a heartbeat.
This will certainly not happen. The reason these places are culture and food deserts is precisely because people drive everywhere and the driving infrastructure requires so much space that it is impossible to have density at the levels needed to support culture.
If you have cheap, abundant land it makes no sense to build densely.
Look at Houston with ~zero zoning laws and ~infinite sprawl.
"A neighborhood" in a high-sprawl suburb wouldn't be able to support local mixed use amenities because even singular "neighborhoods" are gigantic enough to warrant driving across them. Once you're in the car, why would you go to the place 2min down the road instead of the far superior place 8min down the road.
Ask well-paid people who keep renting apartments in Manhattan, or in downtown SF, to say nothing of Tokyo or Seoul.
Ethnic food has thoroughly suburbanized, as has shopping.
Shopping for large items, or large quantities, definitely tends to use suburban land because it's cheaper, and a shopping center uses a lot of it. The cost for the customers is the time to drive there.
They stop at that place and become familiar with it?
How? There would be a huge increase in demand on the roads. You said it yourself, you’d have to build more roads.
Unless you meant, no more [suffering] traffic, since you could just take a nap.
The only way I see self driving to be a true win if it is so efficient that you can remove all the roads and they become part of the mass transit system.
I would demand personal vehicles to pay a premium (cost plus) as they take up more space per person and add to infrastructure maintenance cost
> No sharing space with people on public transit.
If people really want their own private suites they should be paying thru the nose and ears for it. Cars are a worse version of this and the car-centric lifestyle is heavily subsidized by everything from taxes to people's lives (air pollution from ICEs yes, but tire pollution is actually worse in many ways and is made worse with heavier EVs).
This will not fix food deserts, it will make them worse. If your car isn't packed to capacity on every single trip, it is less efficient and worse than public transit.
Roads are awful. We should be trying to minimize them, not expand them.
Whatever ungodly sum you are prepared to pay, I'm certain the actual cost is yet higher.
Then I'll never buy an autonomous vehicle.
I get that most people just want short trips around a major city, but given we, I'm sure it's shocking, don't all live in places like that, or want to spend our time in places like that, it might behoove y'all to solve for other use cases if you want widespread adoption (or at least accept that it's ok to solve for those use cases).
Or, I guess, you can hope that everyone will suddenly decide that all they want is to live in modern Kowloon City because "roads are awful" or whatever memetic nonsense is trending on TikTok.
*In the United States. For reasons we have avoided in much of the rest of the world...
We do have dense pockets. NYC, in particular, has a nice metro (it just needs to be cleaner and more modernized - but it's great otherwise).
Most countries are small. Their dense cities are well-served by public transit. America is just too spread out. Insanely spread out.
China is an exception in that, while a huge landmass, its large cities emerged as the country was wholesale industrializing. It was easy for them to allocate lots of points to infrastructure. And given their unmatched population size and density, it makes a lot of sense.
As much as I envy China's infrastructure (I've been on their metros - they're amazing!), it would be a supreme malinvestment here in the United States to try to follow in their footsteps. The situation we have here is optimal for our density and the preferences of our citizens. (As much as people love to complain about cars, even more people than those that complain really love their cars.)
Public transit in the US is probably going to wind down as autonomous driving picks up the slack. Our road infrastructure is the very best in the world - it's more expansive, comprehensive, and well-maintained than any nation on the planet. We'd be wise to double down. It can turn into a super power once the machines take over driving for us.
The fact that we have this extent of totally unmatched road infrastructure might actually turn out to be hugely advantageous over countries that opted for static, expensive heavy rail. Our system is flexible, last mile, to every address in the country. With multiple routes, re-routes, detours. Roads are America's central nervous system.
Our interstate system is flexible, and when cars turn into IP packets, we'll have the thickest and most flexible infrastructure in the world.
We've shit on cars for the last 15 years under the guise that "strong towns" are correct and that cars are bad. But as it may turn out, these sleeping pieces of infrastructure might actually be the best investment we've ever made.
Going to call this now: in 20 years' time, cars will make America OP.
Those things everything complains about - they'll be America's superpower.
The rest of the world with their heavy rail trains and public transit will be jealous. Our highways will turn into smart logistics corridors that get people and goods P2P at high speed and low cost to every inch of the country.
Roads are truly America's circulatory and nervous system.
I'm so stoked for this. I once fell for the "we need more trains meme" - that was a suboptima anachronism, and our peak will be 100x higher than expensive, inflexible heavy rail.
Autonomous cars that move largely along the same route could form temporary "trains", or rather convoys, moving in a coordinated fashion. That would simplify navigation, reduce chances of accidents, reduce energy consumption, and definitely give the passengers more peace of mind during the commute.
Such convoys would split when needed, join together when needed, notify other convoys and drivers about their route and timing. This would alleviate traffic jams considerably even under heavy load.
At the same time, they would consist of cars and trucks that would be capable of moving completely separately outside highways.
This, of course, will require some kind of centralized control over entire convoys, and a way to coordinate them. Railways and airways definitely can offer examples of how to handle that.
[1] https://www.smartcitiesdive.com/news/fta-transit-ridership-p...
https://www.eea.europa.eu/en/analysis/indicators/share-of-bu...
Will you enthusiastically support the taxes on you needed to entirely offset this negative externality?
Malinvestment into public transit in a way that serves only a limited few of the population and that costs 10x the already high initial estimates is a negative drain on the balance sheet worth 500 billion or more. And this infra is woefully inflexible and static.
California HSR alone is already suboptimal vs. flights, and once we have long distance autonomous self-driving, that'll meet the same demand with 1/100,0000,000th the cost (if you average out the costs and benefits of self driving over all other routes).
I tend to expect better from HN commenters. I don't have an interest in having a discussion with such a callous and dismissive comment. I hope your day gets better.
Tire pollution is worse than tailpipe emissions and the full effects aren't known. You're dismissive of other people's and the environment's health and you're wrong.
https://www.theguardian.com/environment/2022/jun/03/car-tyre...
Tire pollution is now as large or larger than tailpipe particulate pollution, but it’s not a complete apples-to-apples comparison.
Tail pipe pollution includes CO₂, NOx, SO₂, CO, and fine particulates (PM2.5 + PM10) and is strongly linked to asthma, heart disease, climate change.
Tire pollution on the other hand is microplastics, synthetic rubber particles, zinc, and volatile organic compounds. Toxic to aquatic life; long-term human health effects still being studied.
My car is my property. I own it. It does everything I want it to. It is an extension of me.
That question is like asking, "Why own a computer? Why not hire a mathematician to do all your computation for you?"
The problems a self-driving problem solves are 100x deeper than a human, and the second order effects to greater society are enormous. When everyone and everything is self-driving, the roads aren't roads any more - they're TCP/IP and logistics super highways. Anything can go anywhere for any reason at any time. This is a huge societal unlock.
Even thinking about how frictionful ordering an Uber is is exhausting when thinking about the idyllic future of simply jumping into my own car - my own space - and having it do exactly what I want.
This future is magical and I want it now.
I think a fully autonomous car has to be designed around LiDAR and autonomy from the ground up. That's a hugely capital intensive task that integrates a lot of domains and data. And so much money and talent.
This is more in the ballpark of Google Waymo, Amazon Zoox, Tesla/xAI, Rivian, Apple, etc.
And as the other folks have mentioned, this becomes a really good prospect if one company can manage the autonomy, insurance, maintenance, updates, etc. A fully vertically integrated subscription offering on top of specially purposed hardware you either lease or purchase.
This is terrible. In Germany (major city) I pay 166 Eur a month for two cars, one normal (premium brand) family car and second being V8 coupe. I make about 25000km a year in total and have 6 years no claims. No accidents in my driving history (over 15 years). Price is for full coverage with low excess.
You're taking about full coverage, right?
Might as well compare the prices of apples and oranges and vacuums and space stations.
These comments could be quoting liability only insurance or comprehensive/collision for a kia or comprehensive/collision with bodily injury for a rivian R1S. The insured amount would differ by hundreds of thousands of dollars.
For reference, I have only ever paid for maximum liability only insurance including uninsured/underinsured coverage ($500k/$250k), but not bodily injury, and my premium for 10k miles per year is less than $50 per month. Used to be less than $40 per month before 2022.
The medical portion of my insurance that covers me (unlimited PIP) is like $17 a month, I can't see driving much and not spending that, even with relatively limited expectations for how much easier it might make things.
If you can insure the car for less, the car company can charge more for the car. I don't want to pay a subscription (rent) for a car I buy.
And of course there's the fact that you can turn monthly FSD off if you feel the value isn't there. The commitment is much lower, so it's easier to convince people to give it a trial run.
I don't pay for it, though. I still haven't been that impressed with it (we've gotten a couple free months to play around with it). I think in some areas it works pretty well, but in my neighborhood it makes regular attempts to scratch the car.
That's one alternative.
Another alternative would be that you get what you get at purchase time, and you have to buy a new car to get the newest update.
"Continuous development" isn't always a selling point when it's something with your life in its hands. A great example is Tesla. There are plenty of people who are thrilled with the continuous updates and changes to everything, and there are plenty of people that mock Tesla for it. Both groups are large markets that will have companies cater to them.
More likely, one group is a large market that companies will cater to and the other group is a small market that will be very loud about their displeasure on the internet.
Doubt that is a politically tenable model.
"You're telling me my son Bobby died in a crash that could have been prevented with finished software but they only roll it out to people who have the money for a new car despite no technical limitation?" -- yeah, good luck
The Mercedes-Benz model.
We can always choose. The subscriptions aren't mandatory? And there's an alternative to the subscription where they offer it to you for a one time cost.
After all, heated seats are still installed and baked in to the MSRP, even if you're not subscribing to make them work.
The non-Tesla manufacturers have noticed this and positioned products accordingly. Tesla does Musk-driven-development so only caters to the one group.
This site claims the average age of Tesla owners is 48 (updated for 2025):
https://hedgescompany.com/blog/2018/11/tesla-owner-demograph...
Which should not be that surprising. Teslas were priced as premium vehicles initially, and then dropped as competitors appeared and to take advantage of tax credits. Teslas also benefit dramatically from owning a garage and adding a charger to it, which mostly homeowners can do.
A homeowner buying an expensive car is very likely older and richer.
Teslas aren't cool anymore, they are what your parents and your Uber driver has.
Don't get me wrong, as another commenter brought up, I hate traffic too, and the annual fatalities from vehicles are obviously a tragedy. Neither of them motivate me to sign away my rights and autonomy to auto manufacturers.
What happens when these companies decide they suddenly don't like you, cancel your subscription, and suddenly you're not allowed to drive, or I suppose rather use, the vehicle you "own"? It will become the same "subscription to life" dystopian nightmare everything else is becoming.
Or how about how these subscriptions will never be what the consumer actually wants? You'll be forced to pay for useless extra features, ever increasing prices, and planned obsolescence until they've squeezed maximum value out of every single person. I mean imagine trying to work with Comcast to get your "car subscription" sorted.
You know else reduces traffic and fatalities? Allowing workers to actually work from home. Driving during COVID was a dream come true. Let's let the commercial real estate market fail as it was primed to.
People have this strange obsession with over-complicating everything they possibly can.
4) The "smarter" the vehicle, the more they get to track you and sell your data. You'd think "oh in that case I'm sure it'll be like google where I'd pay a reduced price that's offset by the ad money". No, they will obviously happily rip you off on the vehicle itself AND by selling your data. edit: Because guess what? It's working! People are more than happy to fall for this stuff apparently. I mean hell, it's worked for the phone market too, as one other example.
I'd be ecstatic to see the entire industry wiped out by a newcomer on the scene.
Paying subscription for something what they are never going to use is going to be a hard sell.
Corporations could decide to only advertise shipped features, not beta tests.
In America, maybe. Chinese manufacturers are already treating self driving as table stakes. If I have a choice between a subscription car and one that just works, I’m buying the latter.
> continuous development and operations/support
ICE vehicles require continuous servicing and manufacturer support.
Let's revisit this conversation after China's cutthroat automotive competition is resolved. That era passed a long time ago in the US.
Rather than a car which a person may keep for 20 years at a one time huge purchase. Its not really reasonable to expect a company to support something longer than even lifer employees will be there to work on it.
I do really really hate the idea of automotive subscription services though.
Same with Samsung, S25 Ultra ($1100 to $1460) [1], with the fold starting at $2k [2].
[1] S25 Ultra https://www.samsung.com/us/smartphones/galaxy-s25-ultra/buy/...
[2] Z fold, $2000 to $2500 https://www.samsung.com/us/smartphones/galaxy-z-fold7/buy/ga...
At least in cities, a fully-functioning, on-demand autonomous fleet would probably be superior to car ownership in just about every way except as a status symbol.
Meanwhile it's an excuse for another century of more car lanes and less mass transit infrastructure.
Fix those two and personal car ownership will plummet in many places.
Many people don’t want to own a car, pay for insurance, gas, tires, oil changes, parking, washing etc.
Car ownership sucks horribly for most people, it’s just currently the best option. That will change.
I'm sure they could be useful to folks that have the specific use-case for it, but the vast vast majority of trips in a person's day-to-day are better solved by robust multi-modal options and public transit. The benefit there is that less drivers means that traffic is actually better for everyone.
Someone will, I don’t know who. Soon.
How do Waymo prices compare to Uber where Waymo exists?
Some car companies are already trying out subscriptions for stuff that requires zero ongoing support, like seat heaters. Outside of cars, so much software is switching to subscriptions, whether or not it makes sense. The software for my security cameras has become completely infested by ads, but you can pay for a subscription to make them go away. I own the cameras outright, but not really, since the software needed to use them is basically rented, either with cash or with my eyeballs. Most paid apps I come across these days want a monthly fee to keep using it, they're not content to just sell me a copy.
There is a lot of fun in driving a high-hp car on track or offroad or in some not-much-populated area or in plenty of other scenarios. That's where using autonomous driving mode would feel preposterous to me.
And I wish this would be more broadly recognized. Every time there's a story about someone important freaking out about something related to autonomous driving, I'm at least somewhat afraid they'll use it as justification to deny me access to it for those specific use cases.
And honestly, those are the only use cases I really care about or feel comfortable with right now. Of course my car is also too old to support much more than that.
Personally, I look at the 40,000 people killed each year in traffic crashes in the US, and I think, the sooner we all stop driving (on public roads) the better.
It's not a unreasonable cost for development but also maintenance of the self driving system.
> This is a decision. It's generated, I said there's many millions of decisions, many of them will never get noticed and they're just under the surface. One of those decisions that's been noticed quite a bit is the fact that we've intentionally not included CarPlay in the vehicle. And that's not to say we don't think a close partnership with Apple is important. So we have Apple Music integration, we have a bunch of Apple integrations that are yet to come, we have a great relationship with the team at Apple. But it was more to say, we just felt and continue to feel very strongly about creating a consistent, fully integrated digital experience where you're not jumping between apps, let's say from a CarPlay app back to the vehicle app. And it's quite jarring when you don't have, let's say vehicle level controls when you're in the CarPlay environment. That view we've had since the early, early days. I think that's going to become even more important and more true in a world of integrated AI.
They could easily make their screen compatible with Carplay/Android Auto and provide whatever experience Rivian wants to at the same time, and they could let the drivers choose which to use.
And I write this as someone with a Model Y who does not miss Carplay (although it would be nice to have).
What do you mean? My experience in a budget car from the early 2010s is entirely the opposite, so I don't understand fully.
Automotive LiDAR is designed to meet Class-1 laser eye-safety standard, which means "safe under normal conditions." It isn't some subjective/marketing thing, it is an official laser safety classification that is very regulated.
However, if you try to break that "normal conditions" rule by pressing your eyeball directly against an automotive LiDAR sensor for a very long period of time while it is blasting, you might cause yourself some damage.
The reason for why your phone camera would get damaged, but not your eyes, is due to the nature of how camera lenses work. They are designed to gather as much light as possible from a direction and focus it onto a flat, tiny sensor. The same LiDAR beam that is spread out for a large retina can become hyper-concentrated onto a handful of pixels through the camera optics.
Sorry if this is a silly question, I honestly don’t have the greatest understanding of EM.
There is a flip side to this though. Quick searches show that the safety of being absorbed and then dissipated by the water in the eye also makes that wavelength perform worse in rain and fog. I think a scarier concept is a laser that can penetrate through water (remember humans are mostly bags of salt water) which could, maybe, potentially, cause bad effects.
> If you have many lidars around, the beams from each 905 nm lidar will be focused to a different spot on your retina, and you are no worse off than if there was a single lidar. But if there are many 1550 nm lidars around, their beams will have a cumulative effect at heating up your cornea, potentially exceeding the safety threshold.
| Distance | # of Sensors |
| 1 | 1 |
| 3 | 9 |
| 5 | 25 |
| 10 | 100 |
| 25 | 625 |
| 50 | 2500 |
| 100 | 10000 |
It seems like, due to the inverse square law, the main issue is how close you can get your eye to a LIDAR sensor under normal operation, not how many sensors are scattered across the environment. The one exception I can think of is a car that puts multiple LIDAR arrays next to each other (within a foot or two). But maybe I'm misunderstanding something!
Am I being catastrophically pessimistic to think that in addition to swatting insects as it moves forward, the cars lidar is blinding insects in a several hundred meter path ?
I’m very optimistic about automated cars being better than most humans but wonder about side effects.
1550nm might be worse for sensors because a good portion of the light is only being dumped into the metal layers - pure silicon is mostly transparent to 1550nm. Not sure how doped silicon would work. I can tell you that 1070nm barely works on an IQ3 Achromatic back…
Also, it’s something of a nitpick but physically point sources still end up as a circle.
Wasn't sure what level of knowledge you were coming from re: PSFs, so I was keeping it basic.
Class 1 is pretty darned safe, but if you're continually bathed by 50 passing cars an hour while walking on a sidewalk... pitch it to a PhD student you know as something they should find or run a study on.
It isn't however opaque for optical glass (since the LIDAR has to shine through optical glass in the first place) so it hits your camera lens, goes straight through, and slams the sensor.
I was just speaking in terms of the commonplace LIDAR solutions for road use.
There was another discussion a week back https://news.ycombinator.com/item?id=46126780
The lack of accessible certification/testing docs for the lidars is also worrying. Where is the proof that it was even tested? Was it tested just via simulation, via a dummy eye stand-in, or with a real biological substitute?
What if there are biological concerns other than simply peak power involved with shining NIR into the eye? (For instance, it seems deep red light has some (beneficial) biological effects on mitochondria. How do we know that a pulsed NIR laser won't have similar but negative effects, even if it doesn't burn a hole in your retina.)
But yes there are lidar sensors out there where if broken in the right way could burn out your retinas permanently.
Dell and HP don’t make operating systems…it seems like having a handful of companies focused on getting the self-driving part right without the need to also specialize in manufacturing would be beneficial.
My first inclination was to be bullish on Rivian, and there’s no question that their vehicles are beautiful. But is there anything to suggest they have an advantage over Tesla or other automakers when it comes to self-driving?
This has several consequences - Tier 1 suppliers are waiting on input/approval from OEM before proceeding with projects - Tier 1 suppliers don't necessarily have the capital to do work "at risk", even if they could build the part without approval/specifications. (TBH - some do) - Each layer of the supply chain lacks context on the whole project and product line and cannot achieve efficiencies outside of the scope of its contract.
These haven't really been a problem for mechanical parts and E/E parts that have well-defined functions and interfaces and have a lot of re-use from previous generations. It works really well with Kaizen (incremental innovation).
To outsource it, a traditional OEM would need to completely specify the behaviour of said self-driving system, baseline the specification, put out the requests for quotation, etc. Tier 1 then needs to analyse the spec, estimate it, break it down in to sub work packages, work with its suppliers, etc. From an optimisation point of view, this is really inefficient partitioning of the problem space. For greenfields development, an emergent specification via experimentation may be better - but that won't fit in traditional V-model sub-contract OEMs/Tier 1s use.
That flow doesn't need to be followed; the suppliers could raise/allocate capital and build the self-driving stack "at risk" - and this seems to be done (Tier IV, Waymo, etc). But as it's new technology, I assume Rivian think they can do better by themselves and can get the capital for the development as part of an integrated solution while they are smaller it might seem they should not waste limited capitial that way - but integration will save a lot of inefficiency in sharing specifications across boundaries, full system integration and deriving emergent specification via experimentation rather than some MBSE folly.
Real shame nobody has taken that approach, not even a fork
I do wonder if there will be many more iterations, though, with so many manufacturers switching over to an encrypted canbus and locking out the control method comma uses.
These cars will not sell well outside of the US.
I don’t see how Rivian can compete, either on price or features.
But, VW is willing to pay $5B for their software platform. I think they want to extend that to being able to sell custom chips and “AI” capabilities, whatever that means.
* Doors refuse to open
* Lose the ability to control media playback using any controls
* Any button in the UI just opens and closes the windows
Granted, I'm a server side/backend engineer mostly, and I don't know much about writing software/firmware for a very hostile emf environment. But if any project I worked on had bugs like this, fixed at the rate they're fixed on Rivian, I would assume a badly flawed architecture or non existent technical leadership
Yet VW paid billions for this very software. I can't imagine how bad it must've been on their own stack that they gave up and bought this other seemingly broken stack
The Rivian app does not permit you to send a command to the car while the app thinks the car is processing a command. Trunk opening? You can’t unlock the door. On top of this, if you try to open the trunk while outside Bluetooth range and then Bluetooth connects, you are still stuck waiting for the pending command to complete.
Oh, and the ridiculous “hey let’s always remind you that you own a Rivian” Live Activity seems to synchronize on a schedule that involves being hours and hours out of date.
The Rivian app sucks.
One the business side, the economics are fabulous, your competitors can't "clone" your product if they don't have your special sauce components. So in many ways it becomes a strategic advantage to maintaining your market position.
But all of that because the all up cost to go from specification to parts meeting the specification dropped into the range where you could build special parts and still price at the market for your finished product.
A really interesting illustration is to look at disk drive controller boards from the Shugart Associates ST-506 (5MB) drive, to Seagate's current offerings. It is illustrative because disk drives are a product that has been ruthlessly economized because of low margins. The ST-506 is all TTL logic and standard analog parts, and yet current products have semiconductor parts that are made exactly to Seagate's design specs and aren't sold to anyone else.
So to answer your question; apparently the economics work out. The costs associated with designing, testing, and packaging your own silicon appears to be cost effective even on products with exceptionally tight margins, it is likely a clear winner on a product that enjoys the margins that electric vehicles offer.
Yea, Tesla has some. But they aren't sharing their secret sauce. You can't just throw a desktop computer in a car and expect it to survive for the duration. Ford et all aren't anywhere close to having "premium silicon".
So you're only option right now is to build your own. And hope maybe that you can sell/license your designs to others later and make bucks.
They are all expensive, but less than the risk adjusted cost of developing a chip.
And their chips give "1600 sparse INT8 TOPS" vs the Orin's "more than 1,000 INT8 TOPS" -- so comparable enough? And going forward they can tailor it to exactly what they want?
I've been involved with similar efforts on both sides before. Making your own hardware is not a clear cut win even if you hit timelines and performance. I wish them luck (not least because I might need a new job someday), but this is an incredibly difficult space.
NVIDIA also tailor their chips to customers. It's a more scalable platform than their marketing hints at... Not to mention that they also iterate fairly quickly.
So far anyway, being on a specialised architecture is a disadvantage; it's much easier to use the advances that come from research and competitors. Unless you really think that you are ahead of the completion, and can sell some fairly inflexible solution for a while.
In a vacuum there are potentially some advantages to doing your own silicon, especially if your goal is to sell the platform to other automakers as an OEM.
But custom silicon is pricey as hell (if you're doing anything non-trivial, at least), and the payoffs have a long lead time. For a company that's bleeding cash aggressively, with a short runway, to engage in this seems iffy. This sort of move makes a lot more sense if Rivian was an established maker that's cash-flow positive and is looking to cement their long-term lead with free cash flow. Buuuuut they aren't that.
It is unfortunate that the existing market participants seem hell-bent on destroying consumer confidence in self-driving before this will make it to market.
We desperately need a safety regimen for self-driving cars that takes it as seriously as airplane safety.
Also a lot of cars have a lot of limitations with comma.ai. Yes, you can install it on all sorts but there are limitations like: above 32mph, cannot resume from stop, cannot take tight corners, cannot do stop light detection, requires additional car upgrades/features, only known to support model year 2021. Etc.
Rivian supports everything, it has a customer base who LOVE technology, are willing to try new things, and ... have disposable income for a $1k extra gadget.
EDIT: another commenter mentioned that their platform is used by other companies (VW perhaps?). With that information, it might make more sense to do this.