Make your robotaxi predictions for the 8/8 reveal

So Elon says that Tesla will reveal a dedicated robotaxi vehicle on 8/8. What do you think we will see? Will it look like this concept art or something else?

I will say that while this concept drawing looks super cool, I am a bit skeptical if it is practical as a robotaxi. It looks to only have 2 seats which would be fine for 1-2 people who need a ride but would not work for more than 2 people. I feel like that would limit the robotaxis value for a lot of people. Also, it would likely need a steering wheel and pedals for regulatory reasons even if Tesla did achieve eyes-off capability.

So I think this is concept art for a hypothetical 2 seater, cheap Tesla, not a robotaxi.

Could the robotaxi look more like this concept art but smaller? It could look a bit more like say the Zoox vehicle or the Cruise Origin, more futuristic box like shape IMO and seat 5-6 people.

Or maybe the robotaxi will look more like the "model 2" concept:

Other questions:
- Will the robotaxis be available to own by individuals as a personal car or will it strictly be owned by Tesla and only used in a ride-hailing network?
- What will cost be?
- Will it have upgraded hardware? Radar? Lidar? additional compute?
- Will Elon reveal any details on how the ride-hailing network will work?

Thoughts? Let the fun speculation begin!

 

[stopcrazypp]

You literally get fired if you get "strikes" with Waymo, Cruise, or Zoox...
Do you think there is any difference in the safety of testing Waymo software vs. testing FSD software?
I think the DMV is ok with it because there haven't been major issues with safety and Tesla has political sway in California.

Tesla has negative political sway nowadays in California since the move to Texas (plus other Elon things I won't list here as mods may frown upon).
As it relates to FSD, there was a law specifically passed to ban Tesla from using that name.
https://www.sfchronicle.com/tech/article/new-california-law-effectively-bans-tesla-from-17672908.php
Tesla is still fighting a case against DMV trying to ban using FSD naming (as well as Autopilot).
Tesla defends 'Autopilot' and 'FSD' names in false advertising case
There is no way in heck DMV would be giving any leeway to Tesla if they can find a legit reason to force Tesla to get AV permits (that they didn't means they couldn't). But it would be self contradictory given they would be trying to say Tesla FSD is not autonomous thus it can't be called FSD, while at the same time they would be arguing it IS autonomous and it is just under test (thus needing permits).

 

[stopcrazypp]

Optimus will never be sold at meaningful scale. If Tesla delivers 10k Optimus's to real customers this decade will eat my shoe.

BD has plenty of customers like DHL, BP, Maersk, National Grid etc etc

Listen to this (or anyone else that doesn't pump):

Boston Dynamics says robots could be in homes in 10 to 20 years

The personal robots of the future depicted in science fiction could happen a lot sooner in this lifetime than most people think. At Yahoo Finance's Invest summit, Allie Garfinkle interviews Boston Dynamics CEO Robert Playter about the firm's premier robotics projects, "Spot" and "Stretch." Spot...

uk.news.yahoo.com

That's talking about Spot (dog analog) and Stretch (wheeled arm robot), not Atlas (the humanoid robot being compared to Optimus here). There is no commercial application for Atlas yet. In fact the whole recent announcement is them switching to electric motors for the new version because they found the hydraulics they used previously would not be commercially viable. There's still a long way until the new version is ready for commercial release (they are currently only doing R&D with Hyundai, their owner).
Atlas | Boston Dynamics

 

[stopcrazypp]

Because it is? I'm still struggling to see the point of city L2. Doesn't free any time. You're still driving. Highways and traffic jams is another story. There L2 is great.

The same reason people use AP on highways: it frees you from micromanagement of steering/pedals so you can focus on what's on the road.

 

[stopcrazypp]

I wonder what they do after another two-three years down the road they throw in the towel on their Vision only attempt to get them to an L4/L5 (probably even an L3)? Do they ditch the robotaxi stuff and will be overtaken by the Merks and Bimmers as premium EVs with L3 or L4/L5? And won't be able to license their tech because, frankly, nobody will want to license an L2, everybody already has it. Or do they go back to radars/lidars etc like everybody else and play catch-up?

I don't think L3 is worth any effort. It's been more than a year since Mercedes announced L3 certification in the US and there so far is still not a single customer that reported using it yet.
Mercedes-Benz world’s first automotive company to certify SAE Level 3 system for U.S. market | Mercedes-Benz Group
I dug around and it seems at end of last year there were finally some EQS cars optioned with the L3 hardware, but software apparently not activated. It seems even the Mercedes fans think it's a joke to pay $2500 per year for the functionality.
Level 3 Conditional Autonomy Available! - MBWorld.org Forums

Given the low volume of cars with the L3 hardware, plus the fact plenty of people either bought FSD outright or the subscription (even at the previous $199/month), I would venture to guess Tesla has made multiple orders of magnitudes revenue on door-to-door L2 vs other companies have made on L3 (and will continue to do so) .

 

[spacecoin]

The same reason people use AP on highways: it frees you from micromanagement of steering/pedals so you can focus on what's on the road.

If I went to hell have had to work as a driving instructor, I'd much rather be a driving instructor to a student on a highway than in a busy downtown setting. I think almost everyone would agree with that highway driving (with or without driver assistance) is less stressful than driving in a downtown/city setting.

Furthermore, if you would ask a driving instructor if they find it more stressful to sit beside a student or if it's more stressful to drive themselves, I'm pretty sure 100/100 would prefer to drive themselves, all things being equal.

 

[MarkDave]

I predict that on 8/8 Musk will reveal to the world how he plans to synergize FSD with Neuralink. Even if FSD achieves only sub-human reliability, you will be able to regain brain function through Neuralink.

Therefore, FSD will reach mass market adoption way faster than Waymo, which doesn't have any way to revive passengers from failed trips. Only Tesla will be able to revive passengers from failed trips.

FSD + Neuralink = 100% guaranteed survival.

Waymo = no second chances, once the Waymo fails you are injured/dead.

 

[spacecoin]

I predict that on 8/8 Musk will reveal to the world how he plans to synergize FSD with Neuralink. Even if FSD achieves only sub-human reliability, you will be able to regain brain function through Neuralink.

Therefore, FSD will reach mass market adoption way faster than Waymo, which doesn't have any way to revive passengers from failed trips. Only Tesla will be able to revive passengers from failed trips.

FSD + Neuralink = 100% guaranteed survival.

Waymo = no second chances, once the Waymo fails you are injured/dead.

haha!

My prediction for 8/8 is that Musk announces a ”robotaxi” city deployment using Model X with human ”safety” drivers this or early next year. In a jurisdiction without DE and incident reporting requirements. Like the LVCC but with FSDS.

 

[voyager]

Is there truth to Tesla presenting a compact 2-seater AV like this one?

 

[DanCar]

Is there truth to Tesla presenting a compact 2-seater AV like this one?

Speculation might turn into truth, but there is no current truth.

 

[JulienW]

Speculation might turn into truth, but there is no current truth.

...or better facts since truths can be subjective beliefs.

 

[spacecoin]

Is there truth to Tesla presenting a compact 2-seater AV like this one?

They presented the semi and the roadster2 6.5 years ago and neither is in volume production nor has a date set for it.

In other words “presenting” doesn’t mean much. If they present a robotaxi, I expect similar time lines.

 

[JulienW]

One of the things that hasn't been discussed is the data center. Looks like Tesla is building a detected data center for the robotaxie. Another reason our cars aren't likely included in the L4 "moon shot". I bet this data center will take the base code of V12 and then specifically train it on the NEW robotaxie (with likely an *augmented hardware suite). So it will start with the same software but will new compliantly segregated from the consumer cars.

*I predict it will add 4 bumper cameras (for X traffic), a front bumper camera and a rear normal view camera to help it better identify/estimate closing distance high speed approaching traffic.

 

[spacecoin]

One of the things that hasn't been discussed is the data center. Looks like Tesla is building a detected data center for the robotaxie. Another reason our cars aren't likely included in the L4 "moon shot". I bet this data center will take the base code of V12 and then specifically train it on the NEW robotaxie (with likely augmented hardware suite). So it will start with the same software but will new compliantly segregated from the consumer cars.

If they didn’t have a training computer before that explains the slow progress…

 

[stopcrazypp]

If they didn’t have a training computer before that explains the slow progress…

They had one a long time ago. The difference is they are working on one (Dojo) based on an in-house chip (D1, as opposed to buying off the shelf). It's already in operation, but they are using it in parallel with clusters based on Nvidia hardware (as opposed to having all eggs in one basket).
Tesla Dojo - Wikipedia

 

[spacecoin]

They had one a long time ago. The difference is they are working on one (Dojo) based on an in-house chip (D1, as opposed to buying off the shelf). It's already in operation, but they are using it in parallel with clusters based on Nvidia hardware (as opposed to having all eggs in one basket).
Tesla Dojo - Wikipedia

I think the dojo architecture is dead. Why would Tesla buy 10k H100 otherwise?

Turns out nvidia has a few moats with networking and software…

 

[DrChaos]

I think the dojo architecture is dead. Why would Tesla buy 10k H100 otherwise?

Turns out nvidia has a few moats with networking and software…

The moat of moats is access to TSMC best fabs---Apple and nVidia are the worthy customers who can guarantee immense paid volumes and tremendous technical capability. The design at these advanced levels needs to be coordinated intimately with the fab line capabilities and physics and TSMC wants to work only with the customers who will be worth their while and able to keep up.

I suspect Elon is making more and more poorly considered emotional/ego decisions. A Tesla specific on-car inference chip is a good idea: high volumes and opportunities for customization. A Tesla-specific training chip isn't: low volumes and trying to recreate what many other better funded and experienced startups haven't cracked.

Intel bought two startups (Nervana and Habana) for deep learning architectures and that's their core business and they're still struggling.

Most of the other AI chip startups are concentrating on the easy inference side. Accelerating general purpose training is really really difficult while Meta continues to build out pytorch to be optimized foremost for nVidia.

 

[spacecoin]

The moat of moats is access to TSMC best fabs---Apple and nVidia are the worthy customers who can guarantee immense paid volumes and tremendous technical capability. The design at these advanced levels needs to be coordinated intimately with the fab line capabilities and physics and TSMC wants to work only with the customers who will be worth their while and able to keep up.

I think CUDA and building a scalable software stack of super computers and other infra is as a big moat.

 

[aataskin]

I don't think L3 is worth any effort. It's been more than a year since Mercedes announced L3 certification in the US and there so far is still not a single customer that reported using it yet.
Mercedes-Benz world’s first automotive company to certify SAE Level 3 system for U.S. market | Mercedes-Benz Group
I dug around and it seems at end of last year there were finally some EQS cars optioned with the L3 hardware, but software apparently not activated. It seems even the Mercedes fans think it's a joke to pay $2500 per year for the functionality.
Level 3 Conditional Autonomy Available! - MBWorld.org Forums

Given the low volume of cars with the L3 hardware, plus the fact plenty of people either bought FSD outright or the subscription (even at the previous $199/month), I would venture to guess Tesla has made multiple orders of magnitudes revenue on door-to-door L2 vs other companies have made on L3 (and will continue to do so) .

I wonder how many of them knew they were only buying a "door to door" L2.

 

[JulienW]

I wonder how many of them knew they were only buying a "door to door" L2.

I doubt that many more than 1% has even a redemptory understanding of SEA J3016.

 

[MarkDave]

I wonder how many of them knew they were only buying a "door to door" L2.

One person said it best about FSD:

"If you are disappointed by $TSLA EV adoption, wait until you see AV/FSD adoption."