ra7

Volvo had contract issues with Luminar:

> In a statement, a Volvo Cars USA spokesperson added the decision was made “to limit the company’s supply chain risk exposure, and it is a direct result of Luminar’s failure to meet its contractual obligations to Volvo Cars.”

Groq calls their inference chips “Language Processing Unit”: https://groq.com/blog/the-groq-lpu-explained

Highly ironic considering you started this comment chain with a bunch of fanboy talking points and misinformation. Clearly, you’re not interested in being factual. Bye.

> More isn't objectively better, less isn't objectively better.

Great, you finally got there. All it took was one round of correcting misinformation about LiDAR and another round of completely useless back-and-forth about sensor count.

The words you’re looking for are necessary and sufficient. Cameras are necessary, but not sufficient.

> There's only different strategies and actual real world outcomes.

Thanks for making my point. Actual real world outcomes are exactly what matter: 125M+ fully autonomous miles versus 0 fully autonomous miles.

Maybe Tesla should reduce their camera count from 8 to 2 and put them on a swivel like human eyes. Less is surely always better.

I can also make “clever” arguments that are useless.

It's just "sensing depth" the same way cameras provide just "pixels". A fused cameras+radars+lidar input provides more robust coverage in a variety of conditions.

> The only thing LIDAR can do sense depth

This is absolutely false. LiDAR is used heavily in object detection. There’s plenty of literature on this. Here’s a few from Waymo:

https://waymo.com/research/streaming-object-detection-for-3-...

https://waymo.com/research/lef-late-to-early-temporal-fusion...

https://waymo.com/research/3d-human-keypoints-estimation-fro...

In fact, LiDAR is a key component for detecting pedestrian keypoints and pose estimation. See https://waymo.com/blog/2022/02/utilizing-key-point-and-pose-...

Here’s an actual example of LiDAR picking up people in the dark well before cameras: https://www.reddit.com/r/waymo/s/U8eq8BEaGA

Not to mention they’re also highly critical for simulation.

> It can't read road signs. It can't read road lines.

Also false. Here’s Waymo’s 5th-gen LiDAR raw point clouds that can even read a logo on a semi truck: https://youtube.com/watch?v=COgEQuqTAug&t=11600s

It seems you’re misinformed about how this sensor is used. The point clouds (plus camera and radar data) are all fed to the models for detection. That makes their detectors much more robust in different lighting and weather conditions than cameras alone.

Tesla literally has a human in the driver seat for each and every mile. Their robotaxi which operates on geofenced “guardrails” has a human in the driver seat or passenger seat depending on area of its operation, and also has active remote supervision. That’s direct supervision 100% of the time. It is in no way similar in capability to Waymo.

We’ve been hearing Tesla will “surpass Waymo in the next 1-2 years” from the past 8 years, yet they are nowhere close. It’s always future tense with Tesla and never about the current state.

Correct.

Tesla can’t go on highway (or anywhere) without a safety driver or a safety monitor in the passenger seat after nearly a decade of development. Not sure I’d call that “way ahead”.

Can Tesla make its software autonomous enough to match Waymo?

There’s universal agreement that DRS has massively improved cricket, but the sport was already using technology for decision-making well before DRS. In fact, Hawk-Eye was first used in cricket before gaining popularity with tennis. I’d go as far as to say cricket has been the best adopter of technology in sports for a long time.

In a recent interview, Waymo strongly hinted they now have positive unit economics in markets like SF.

From https://www.forbes.com/sites/alanohnsman/2025/09/03/waymo-co...:

> “Each car, the amount of revenue it's making would be shocking to most people,” Panigrahi said, without elaborating. “Because it just continuously keeps delivering ride after ride. On a per asset basis, it’s doing really well. That’s making progress in terms of unit economics very, very positive.”

So much so that in busy markets like San Francisco, Waymo could soon move into the black. “Not making specific statements about if we’re positive or not, but what I can tell you is that yes, key markets are showing us that we are,” Panigrahi said.

Yes, but SFO (and the city of SF) has a long history of being hostile towards Waymo: https://missionlocal.org/2023/11/waymo-rebuffed-by-sfo-sf-gu...

What likely happened now is that SFO got a kick up their backside from the Mayor after the press started asking why it was still dragging its feet, while SJC approved Waymo swiftly.

Probably because SFO felt the heat after Waymo acquired SJC approval quickly: https://sfstandard.com/2025/09/05/phoenix-has-airport-robota...

They already do in PHX.

Tesla robotaxis have teleoperation [1], which is worse than remote assistance others use because the operators have direct control. Yet they cannot fully remove safety personnel from the car.

The old ambition is dead.

[1] https://electrek.co/2025/05/16/tesla-robotaxi-fleet-powered-...

$200 to enable better FSD vs a decade of struggle to get FSD only partially working with $20 cameras. Which one do you think is more expensive overall?

The complexity argument rings hollow to me. It’s a bit like saying distributed databases are complex because you have to deal with CAP guarantees. Yes, but people still develop them because it has real benefits.

It was maybe a valid argument 10 years ago, but in 2025 many companies have shown sensor fusion works just fine. I mean, Waymo has clocked 100M+ miles, so it works. The AV industry has moved on to more interesting problems, while Tesla and Musk are still stuck in the past arguing about sensor choices.

New York is a long way behind California in regulating autonomous vehicles. Fully driverless vehicles are also illegal there and it will require legislation for Waymo to deploy in the state.

> “Tesla designed Autopilot only for controlled access highways yet deliberately chose not to restrict drivers from using it elsewhere, alongside Elon Musk telling the world Autopilot drove better than humans,” Brett Schreiber, counsel for the plaintiffs, said in an e-mailed statement on Friday.

Tesla had every opportunity to do the safe thing and limit Autopilot to only highways, but they want the hype of a product that "works anywhere" and don't want to be seen as offering a "limited" product like their competitors. Then they are surprised their users misuse it and blame it all on the driver. Tesla wants to have their cake and eat it too.

You know what else prevents misuse? Implementing safeguards for your safety critical product. It took multiple NHTSA probes for Tesla to implement obvious driver monitoring improvements to Autopilot/FSD. There's a reason Tesla is always in hot water: they simply lack the safety culture to do the right thing proactively.

What I gathered from comments here is it's not a simple flick of the switch and it actually takes some effort to turn them off. Can you really do it twice within the span of 1 second?

Context is both these switches being turned off with a 1 second gap. Doing it with one hand simultaneously would possibly explain it, otherwise it doesn’t seem relevant.

It isn’t a pivot. Rather it’s gaining business that would otherwise go to Broadcom/Marvell. They continue to keep their high margin proprietary product. They don’t have to give away all their design secrets, just enough to make them a valuable chip design partner.

Admittedly, I haven’t thought this out, but I can’t find compelling reasons why they shouldn’t do this.

There's a lot of push for inference hardware now (e.g. Ironwood TPUs). How does Nvidia maintain an edge there?

Also, I think the market has to expand beyond LLMs to areas like robotics and self driving cars (and they need to have real success) for Nvidia to maintain this valuation. I don't think only LLMs are enough because I don't see code assist/image generation/chatbots as a massive market.

I understand Nvidia is in a very dominant position. But $4T market cap still seems absolutely insane to me. I've only read about the Cisco boom and bust during the Internet era, and this feels eerily similar (people who actually experienced it might feel differently though).

What could actually drag Nvidia down and make them spend decades in the dark like Cisco still does? So far the two things I've come up with are: (a) general disillusionment in AI and companies not being able to monetize enough to justify spending on GPUs. (b) Big companies designing their own chips in-house lowering demand for Nvidia GPUs.

I don't think Nvidia can counter (a), but can they overcome (b) by also offering custom chip design services instead of insisting on selling a proprietary AI stack?

The desired package management system they describe sounds a lot like Carvel's kapp-controller (https://carvel.dev/kapp-controller/). The Carvel ecosystem, which includes its own YAML templating tool called 'ytt', isn't the most user friendly in my experience and can feel a bit over-engineered. But it does get the idea of Kubernetes-native package management using CRDs mostly right.

This was a recent change by the current administration to loosen previously stricter data reporting requirements: https://www.reuters.com/business/autos-transportation/us-age...

No serious analysis can be done when we can’t even tell if a crash occurred under FSD Supervised or Autopilot because they’re two very different things with different capabilities. Same with withholding software/hardware versions and narrative of events.

Tesla also has a problem of their telematics underreporting crashes. One of the reasons for that is they don’t consider it a crash if airbags don’t deploy. This was called out by the NHTSA in a prior investigation: https://static.nhtsa.gov/odi/inv/2022/INCR-EA22002-14496.pdf.

Here’s the relevant paragraph from that report:

> Gaps in Tesla's telematic data create uncertainty regarding the actual rate at which vehicles operating with Autopilot engaged are involved in crashes. Tesla is not aware of every crash involving Autopilot even for severe crashes because of gaps in telematic reporting. Tesla receives telematic data from its vehicles, when appropriate cellular connectivity exists and the antenna is not damaged during a crash, that support both crash notification and aggregation of fleet vehicle mileage. Tesla largely receives data for crashes only with pyrotechnic deployment, which are a minority of police reported crashes.

Crash data for all other ADAS systems is already public [1]. The only manufacturer with heavily redacted information in that data to the point of being useless is Tesla.

[1] https://www.nhtsa.gov/laws-regulations/standing-general-orde...