Hacker News Reader

Comment by TheAlchemist | Hacker News Reader

So, does that mean that all the Tesla on the roads won't become robotaxis with a simple over the air update ? But at least, Optimus is still scheduled to start serious work in factory right ? /s

On a more serious note, I don't get why people think robotaxi have such a huge business potential. Robotaxi are certainly the future - that's undenialable. But is it such a good business ?

I don't see many people ditching their cars for a Robotaxi. Those who would, don't own a car anwyay. As for margins, it seems to me like it will be a race to 0, quite like airlines, with very little profit to be really made.

View on HN

I think at an even smaller scope than robotaxis, having a really convincing level 4 driver assist (dare I say "full self driving" mode) for general consumer vehicles is a decent value proposition. Both for prospective new car buyers and existing owners. EV adoption has still not reached saturation, so giving some compelling reason to buy a Tesla over another car is pretty big.

After a car rolls off the lot its hard to make money on it. Tesla is in a position to actually make money off this, provided their FSD offering is compelling enough. You can extract decently high margin recurring revenue from potential FSD subscriptions.

And despite their shortcomings, I think they are probably the closest auto OEM to making it happen, and they are clearly taking things pretty seriously (multi billion dollar GPU clusters, paying top dollar for talent, having a massive data edge, etc).

I think Waymo is the superior technology as of now, but in the short - medium term, having a compelling FSD product on a consumer car will probably benefit Tesla quite a lot. I'm not too sure how compelling a fully car-less (i.e people only commute via some company's robotaxi fleet) future is in the medium term though, and its business viability, but it seems like a not-unreasonable north star to work towards.

Do you think that there exists a big-enough market of consumers who will happily pay a recurring subscription for FSD-esque features in their car?

What would be the price point of such a subscription? I struggle to think of a cost that strikes the balance between value for money and me being suspicious of the quality of the service I'm getting. 100s of dollars could easily double or triple the monthly lease/loan which will directly impact the value-prop math we do do in our heads when buying a car. Tens of dollars will immediately make me suspicious of the quality of the service I'm getting.

I'm not so sure about any of this. My feeling is that the majority of consumers will not opt of such a subscription, especially if it's something "I can just do myself for free" (drive my car). I have no data to back this feeling though, so I'm curious to know what others think.

What do you even subscribe to? Is the alleged self driving run by a data center somewhere?

That was going to be my follow on question.

Presumably the answer is a bunch of tele-operators in a 3rd world country, based on this article.

Margins would only be low if there are multiple companies with similarly good software.

But that is usually not what happens in the software space, especially for software that is expensive to build. Google still has its lead. About $200B in search revenue per year and still growing. Google's overall margin is 32% and the search business probably has an even higher margin.

Tesla has a good shot at becoming the first company to have a self-driving software that can run on vision alone.

Come on, it's time to ditch this "vision alone" thing for self-driving cars. A computer will never be safe with vision only. Things like reflection and false positives will always wreck the statistics of even the perfect AI, there is no way around it. And I'm talking about good weather conditions.

The numbers show it, the few that Tesla releases about FSD are not up to the level with Google. If it was good, Tesla would be transparent with their results.

I think vision some will likely eventually be sufficient for full self driving, but by the time it is vision and LIDAR will be cheap and ubiquitous and downgrading to vision alone will be seen as a cost cutting measure that unnecessarily degrades safety compared to systems with LIDAR as well.

Then how do humans do it?

Human have excellent 3d vision, even with one eye. Some ability that only a vision-lidar fusion would match, not vision only. Human vision isn't just a flat image like a camera render. Distance from an object is not only estimated from using a stereo camera (2 eyes), it is made from constant movemnt from the head, and the ability to rotate the orbit on two axis.

Vision might be viable in an unknown future (which tesla is very good at seliing), but right now it is very dumb to forbid yourself extra data.

    ability that only a vision-lidar fusion
    would match, not vision only

What would be arguments that back this theory up? Why couldn't an image stream from fixed cameras be enough?

After all, we made machines perform on a superhuman level in many areas already. You could have said "A machine will never be able to move as fast as a human unless it has two legs and feet". But that is not how it turned out.

I remember people saying computers will never beat humans in chess because they don't have intuition.

I don’t see any reason to argue possibility- it’s more interesting to know what’s optimal. Would you agree that redundancy in sensors is better?

Optimal towards which metric?

What i meant is that if you multiply the number of cameras, probably make them movable, and with amazing vision software you might approach the performance of 2 humans eyes, but really it is just extra step to simulate lidar data at this point.

> After all, we made machines perform on a superhuman level in many areas already. You could have said "A machine will never be able to move as fast as a human unless it has two legs and feet". But that is not how it turned out.

Yes, thanks to wheels. Robots with legs do exist; but none of them is close the speed of an average human running. That's my point. We are not gonna do better than human vision by simply taking 2 cameras thinking it is equivalent to a human eye.

How do you know that taking 2 cameras and superhuman intelligence won't be enough?

Because it's physically not the same. Again, 1 fixed camera does not provide as much data as a single eye. It's not about the resolution, it's about the 3d perception that even a single eye provides. You can try yourself : close one eye, and you still have a good 3d perception. Now try to make your orbit fixed, no head or eye movement, and then everything feels flatter.

Metal wings are also not physically the same as two feet. Yet, planes can move faster than humans.

How well cars will be able to maneuver with cameras and superhuman intelligence specialized in this task is yet to be seen.

Maybe we could try to find dashcam videos of self-driving cars causing accidents. Then one could form an opinion if the camera footage will be enough for future versions of the software to correctly assess what is on it.

Here is one:

https://www.youtube.com/watch?v=Jcj-1hNwPiw

The crash is at 1:08

To me, it seems entirely possible to understand the situation from the camera footage alone.

By being the greatest general intelligences we know of? Even then, it’s still the most lethal thing most people do.

If we get a human level AGI with vision as high dynamic range as ours then maybe we can match human driving performance.

Seems easier to use radar and lidar at that point!

How birds fly? Well definitely not like planes.

Sometimes it is easier, far simpler and even superior not to mimic nature. Consider if it is even possible to have supersonic flight by mimicking flapping of wings.

They don’t do it well. 100 people will be killed TODAY on roads in the US. (And everyday)

Humans do it badly. Humans do it by killing 50,000 other humans per year in the us alone. A million worldwide.

Their hardware has more memory, compute capability and throughput.

Also the image from the retina is higher quality compared to primitive cameras.

What is the point of making pretensious snob comments like this?

You are out of touch, less young people have their licenses than ever before. Car ownership, maintenance, and insurance are expensive. There is a huge market for people who would rather hail a robotaxi than buy a car. Companies like uber and lyft wouldn't be generating billions in revenue if that weren't the case.

That's my question - Uber / Lyft generate billions (low tens, not hundreds!) in revenue, and pretty much no profits at all. It doesn't seem like an excellent business at all.

Have you updated your priors? Uber is profitable today.

> Car ownership, maintenance, and insurance are expensive.

Not as expensive as uber/lyft.... by a fairly large margin, too. The kids are using public transit.

> There is a huge market for people who would rather hail a robotaxi than buy a car.

? Do they have hailing? I thought you would have to use an app. If I can actually hail a robotaxi I might use one

> ? Do they have hailing? I thought you would have to use an app.

I´d call this an example of language evolution comparable to 'calling someone' which used to involve either raising your voice so the person could hear you or walking towards wherever that person is but now involved pulling out a communication device which bridges the gap between caller and callee. The same goes for 'hailing a cab' which used to involve standing on the side of the road, hand raised and possible shouting 'taxi' upon which a horse-drawn or self-propelled vehicle with driver pulls up to drive you somewhere in return for payment but now involves pulling out a communications device which bridges the gap between customer and 'driver'.

It's a great business. It's like Uber (which is worth $150 billion), but with 3x less costs. Taxi services are network effects that are extremely difficult to compete with, which is why the taxi industry is basically a duopoly now.

How will operating cost be 3x less compared to Uber?

Uber drivers are already low-wage earners in many instances, moving them all to a tele operator centre in a 3rd world country will not make that much of a difference I think. I also suspect the cars will be significantly more expensive compared to the typical Corrola or similar shitbox that Uber drivers prefer. And this cost won't be foisted on the driver, and will be a direct capex for Robotaxi. Charging, charging infra, maintenance and all that will also be a running cost that Uber does not have deal with.

How exactly is it 3x less costs? Uber drivers' margin is extremely low, it varies a bit between countries but after energy costs, maintenance/upkeep costs, vehicle acquisition cost and depreciation, etc. their take home pay is low. Pluck the driver out and you still have a car that needs maintenance and energy to run and savings are on the labour cost which Uber already pushed down as much as possible.

I don't see anywhere a way to get 3x less costs in this equation.

Uber barely makes any money actually, how is it a great business ?

Valuations are completely meaningless currently.

FYI Uber reported a net income of $2.6B in Q3-2024 and $1.9B for 2023. That's no Google but that's also no "barely makes any money" territory, no?

Source: https://www.google.com/finance/quote/UBER:NYSE?hl=en

That's no Google but that's also no "barely makes any money" territory, no?

Yes.

Income and profit are not the same.

Uber spends all their income and then some --- which actually leaves them in "makes less than no money territory".

Imagine winning $100 at a casino. You tell your buddies about the big win but you conveniently neglect to mention the fact that you lost $200 first.

Your income was $100 but your expenses were $200 which leaves you with a net loss of $100.

Income is profit. Uber made $2.6 billion.

What you are talking about is revenue.Uber generated $11.2 billion in revenue.

Their profit margin was 23%. Up from 2.3% a year ago.

Business terminology is precise. Any founder that makes this mistake in front of an investor will lose their trust. Know the difference.

eh... I'm pretty sure I mentioned net income (which i think is pretty close to profit in some circles) but if we care about debt servicing, buybacks...etc, let's just look at cash flow. And there too, in Q3, they posted $1B+ for quarter.

I'm not sure I understand the point of your analogy with respect to wining/losing at a casino. If I follow it, you are effectively saying something akin to, when Google first became profitable, its achievements should have been completely dismissed because of a decade of prior losses. That's not how tech companies work. Most of the value creation is 5-10-20 years out. You bear enormous losses with the knowledge that returns compound and eventually render irrelevant prior years' losses.

Yeah, and also a loss of $8.5B in 2019, loss $6.7B in 2020 and loss of $9.14B in 2022. Hell of a business they got there.

So? I can make the exact same point about Google and other tech companies that endured years—decades sometimes—of eye-watering losses, only to eventually build enormously valuable businesses.

Of course. But it's a bit premature to say it's a good business and making money.

It may or may not become a good business.

Let's say Tesla does manage to launch a cheap Robotaxi, fully autonomous, next year ? (Of course they won't, they are decades away from that)

Uber would be pretty much dead, and all it would have done in it's lifetime is massive losses.

Uber is profitable with strong FCF.

https://investor.uber.com/news-events/news/press-release-det...

How does it compare to a car sharing company? It has all the same costs for the vehicle and the driver/pilot is free in both models, in one it is AI and the other it is just the user.

You don't need a driver's license.

It's much higher cost. Currently drivers own the cars and therefore taxi companies get that capital for free. Self-driving taxi companies will have huge capital costs because they'll have to buy (or lease) the cars.

you just spit numbers from your ass don't you

People who live and/or work in city center where it's near-impossible to park. If the robocab could pick me up, drop me off and find the parking space outside the city center – that'd certainly be a great product.

How is this better than current Uber/Taxicab services? Or the bus, or the tram, or the metro, or a rental ebike/scooter?

Not having a stranger in your car is a big benefit for one.

Usually I'm the stranger in someone else's car when I'm taking a taxi but idk

You're a customer and it's their choice to have you in their car.

I can only imagine how confused you must be by "your taxi arrived" notification

It only needs to be cheaper.

It beats bus, tram and metro on convenience.

It beats ebikes and scooters on convenience and security.

The vision for robotaxis at this stage it pretty obvious and like you say undeniable - the profits though will be mostly there for those with first mover advantage that can make their brand name the verb for driving somewhere.

It didn't take great vision or foresight to see the uber app back in the early 2010's and see how things will play out vis-a-vis getting rid of the driver - and my money is still with them (literaly) from a brand name perspective only.

What does Uber have? Except exceptional losses in the 50B range? Like do they have any hardware at all?

The entire Selfdriving/Robotaxi space goes the same way that ASIC miners for bitcoin did - the producers find it most profitable to run it themselves while selling expensive hardware to end users.

> It didn't take great vision or foresight to see the uber app back in the early 2010's and see how things will play out vis-a-vis getting rid of the driver

It's not clear to me. The Uber we have today seems to be paying the third party drivers around cost price per kilometer, and doesn't have to allocate capital for purchasing vehicles, nor for cleaning or maintaining those vehicles. An Uber that had to buy a bunch of self-driving vehicles that they would also then have to maintain, clean, and service, seems like a worse, more risky, business than what they are running today.

They will surely find a way to make that into a "gig job" too. Like the guys who currently pick up e-scooters, drive them away using various conveyances, charge them and drop them off again - maybe robotaxis can drive themselves to a charging station, but cleaning them will surely still be a gig job.

> The Uber we have today seems to be paying the third party drivers around cost price per kilometer

That doesn't make sense, as the drivers are obviously not operating at a loss.

Wait why "obviously"? Is this like how "obviously" the friends you have in an MLM scam must be making money? Or "obviously" Wall Street wasn't packaging dogshit "no job, no proof of income" mortgages as triple-A securities before the crash?

If by "obviously" you only mean "That seems dumb" then, yeah, welcome to planet Earth.

I own a car which sits unused for 99% of the time. Traditional taxis and unreliable/unpleasant public transport are the reason I keep it.

I can imagine many people are in the same position and would gladly do without the hassle if only there was an easy and reliable alternative.

The easy, reliable and cheap (aka practical) alternative would probably be a small EV like this one:

https://electrifynews.com/news/auto/chinas-new-10000-ev-byd-...

But there is a long list of social, economic and political forces determined to prevent this from ever happening in the USA.

"quite like airlines"

With lower entry barriers on top.

It could also be that Musk thinks it's a market where he can avoid liabilities.

It's going to trigger a race to the bottom I think. Robotaxi is going to run into exactly the same challenges that Uber experienced back in the day: significant push-back from existing taxy industries.

In markets where they are allowed to operate they will compete directly with Uber which will probably drive down initial cost (consumers win), but the inevitable enshittification will no doubt take place, just making both options terrible.

The existing taxi industry has largely been crushed by Uber, while Uber would love nothing more than to get rid of pesky "driver-partners" who have the effrontery to demand annoying and expensive things like fair working conditions and health insurance. And this isn't just speculation, Uber has already announced a partnership with Waymo rolling out from 2025.

> The existing taxi industry has largely been crushed by Uber

I have limited experience, but my two data points:

In South Africa Uber and the existing taxi industry operate on two completely separate and non-mixing markets. Even though Uber (and similar services) is getting a lot of abuse from the taxi industry, I don't think the taxi industry has really been affected by Uber at all.

In Ireland Uber is basically non-existent. Only sanctioned taxi drivers are allowed to be drivers on Uber. So they are basically just regular taxi drivers who happen to be drivers on the Uber app. It's actually really difficult to get an Uber where I am, there's only like one or two of the taxi drivers who are actually on the app.

I have extensive experience, and the only other country I can think of where ridesharing apps are straight up banned is Mongolia. The last major holdout was Japan, which until recently only allowed licensed taxis to be hailed, but they've started experimenting with liberalization as well.

Uber's own stats are 9.4 billion trips across 161 million users in 2023, and this excludes those markets where it has withdrawn in favor of other rideshare providers (Didi, Bolt, Grab, Yandex, Careem, etc).

Reply on news.ycombinator.com