Blog Tesla May License Self-Driving Software to Other Automakers

Tesla may license its self-driving software to other automakers. In fact, Chief Executive Elon Musk said the company has had “preliminary” discussions with potential partners.

Musk mentioned the meetings during Tesla’s Q4 2020 earnings call.

“We’ve had some preliminary discussions about licensing Autopilot to other OEMs,” Musk said on the call. “This is something we’re more than happy to do.”

Musk noted that for such a partnership to take place, Tesla would need to further prove that Autopilot is safe in full self-driving mode. Musk said the proof could come by the end of the year.

Musk didn’t disclose which other automakers have expressed interest in licensing the technology. While others are developing their own in-house technology, Tesla’s fleet has logged billions of miles to train the software. The company may have a big enough head start that it makes sense for the industry to adopt Tesla’s offering.

 

[diplomat33]

I think licensing FSD to other automakers could be very tricky.

For one, Tesla would need to get FSD to be super safe and reliable first. Other automakers won't want to license something if they think it is not safe enough. Right now, FSD is still very much beta and not true autonomous yet.

Second, most other automakers want other sensors like a driver facing camera to monitor driver attention and lidar. Tesla does not believe in either. So how will Tesla license FSD if the hardware is not compatible? There could issues with training the camera vision to work for other automakers that have different cameras or place the cameras differently on the car. So it might not be an easy "plug and play" solution.

Thirdly, most other automakers have their own research on FSD or have partnerships with AV companies. They won't need Tesla's FSD if they have something that is as good or better.

 

[bobleland]

I think this is very smart and should be the industry standard. It greatly simplifies the future safety to have all FSD vehicles singing out of the same song book. I would like to see it as mandatory as long as there is room for individual OEM's to add on subroutines to their fleet. .

 

[jeremymc7]

I don't see too many, if any of the big boys doing this. They'll likely want to roll their own, cooperate with each other, or buy some other third party system from a non auto maker that sells parts / systems to multiple auto makers.

 

[Zimo]

If you're an OEM and you know you are many years behind in developing this software, it would make sense to license Tesla's. If you're the first to do this, you have a big sales advantage. This would then force others to consider this option as well.

 

[Bladerskb]

If you're an OEM and you know you are many years behind in developing this software, it would make sense to license Tesla's. If you're the first to do this, you have a big sales advantage. This would then force others to consider this option as well.

The question is, do these OEMs have options, are those options available right now and what do they cost compared to Tesla?

 

[Cheburashka]

If you're an OEM and you know you are many years behind in developing this software, it would make sense to license Tesla's. If you're the first to do this, you have a big sales advantage. This would then force others to consider this option as well.

I doubt this is the case.

GM has Super Cruise in production vehicles and has Cruise FSD in testing, for example. The other manufacturers all have similar deals.

I really don't see the purpose of licensing Tesla's "FSD" considering how unreliable and flaky it is in the current form.

 

[mikes_fsd]

I doubt this is the case.

GM has Super Cruise in production vehicles and has Cruise FSD in testing, for example. The other manufacturers all have similar deals.

I really don't see the purpose of licensing Tesla's "FSD" considering how unreliable and flaky it is in the current form.

This is not how this will happen.
I think it will start with the regulators wanting to require active safety features...
Safety features will start being rated as part of crash testing (probably in the next 5 years) and safety standards will be expanded to have requirements (like the backup camera is now required on all cars built since May 2018) if any manufacturer cannot provide the active safety features, they will license from someone like Tesla. But once you license those features (which will require the FSD computer) you might as well offer (and mark up) the entire features stack to your buyers.

 

[Bladerskb]

This is not how this will happen.
I think it will start with the regulators wanting to require active safety features...
Safety features will start being rated as part of crash testing (probably in the next 5 years) and safety standards will be expanded to have requirements (like the backup camera is now required on all cars built since May 2018)

We already see how slow regulations are. Infact regulation for 2025 is already laid out (Euroncap 2025) and it doesn't include alot of things Safety Group have been calling for.
Guess what's one of the requirements? Driver monitoring. Yes something Tesla's don't have but alot of automakers have now. It will actually be required from 2022 i believe.

if any manufacturer cannot provide the active safety features, they will license from someone like Tesla. But once you license those features (which will require the FSD computer) you might as well offer (and mark up) the entire features stack to your buyers.

All OEMs already have active safety features as an option and they all meet the requirements from regulation. No need to license from Tesla.
Also it doesn't require a 114 TOPs "FSD computer". OEMS get it dirt cheap. Another reason why they won't get it from Tesla.
when they can use a $10 or $45 Mobileye chip versus paying $5,000-10,000.

NHTSA Announces Update to Historic AEB Commitment by 20 Automakers

 

[Cheburashka]

This is not how this will happen.
I think it will start with the regulators wanting to require active safety features...

If this is the case wouldnt Tesla have to license things from other manufacturers, not the other way around?

Teslas don't have basic safety features like rear cross traffic alert that even econoboxes have these days.

 

[voyager]

At the AI Day event, Autocar asked Musk if Tesla would consider making their FSD software open-source, to which he replied: “Well, it is fundamentally extremely expensive to create the system, so somehow that has to be paid for. Unless people want to work for free. But I should say that, if other car companies want to license it and use it in their cars, that’d be cool. This is not intended to just be limited to Tesla cars.”

Tesla Could Allow Other Automakers To Utilize Its Self-Driving Technology | Carscoops

 

[diplomat33]

At the AI Day event, Autocar asked Musk if Tesla would consider making their FSD software open-source, to which he replied: “Well, it is fundamentally extremely expensive to create the system, so somehow that has to be paid for. Unless people want to work for free. But I should say that, if other car companies want to license it and use it in their cars, that’d be cool. This is not intended to just be limited to Tesla cars.”

Tesla Could Allow Other Automakers To Utilize Its Self-Driving Technology | Carscoops

Licensing FSD to other automakers could potentially be a lucrative source of income. However, as I mentioned up thread, there are some issues:
1) Automakers would need to install their own hardware. Presumably, it would need to be a similar 8 camera set-up and a FSD computer like Tesla has done. That would cost money.
2) If automakers have a different set-up, how easy would it be to adapt the FSD software? Who would be responsible for doing that adaptation and making sure the software works? And if an automaker does license the FSD software and there is an accident, would Tesla be liable? And some automakers do want to use radar and lidar. Tesla's FSD would not work for them.
4) Many major automakers already have their own ADAS/FSD strategies, either doing their driver assist in-house or outsourcing it. So they might not be interested in Tesla's product. For example, GM does their own ADAS and has Cruise that is doing FSD. Ford has ArgoAI for FSD. Mercedes is doing L3 in-house.
5) Tesla's software is still beta. Right now, I doubt that many automakers will want to license beta software. Tesla would need to finish the software and validate that it is safe before they could license it to others. Furthermore, Tesla has a bad reputation when it comes to safety with some high profile accidents in the news. Heck, Tesla is under investigation right now. So I think other automakers would be unlikely to want Tesla's software right now. Tesla would need to really prove that it is safe enough first.

 

[RDoc]

Wouldn't other car makers want to license software that works and was maintained by a reputable company?

 

[Dan D.]

Wouldn't other car makers want to license software that works and was maintained by a reputable company?

In the short term, yes I can see some car makers partnering up, rather than waste R&D money.

I'll take it another level. There may come a time where you have to use only approved systems. It may become very cumbersome to deal with re-regulation (like with the airline industry), and only large AV makers can cope. There may only be a limited choice, or perhaps some arm of the government owns/licenses them.

We're not at this point in regulation, but I can see it happening one day - down the road. For now, we're well off from even knowing what kind of system works best.

 

[kaffine]

The only car companies that I could see wanting to license FSD in its current form would be the other upstart BEV companies.

Any of the major auto companies are going to want a more reliable system even if it has less abilities. For those major auto companies that don't have internal R&D programs for this will go with one of the major suppliers offerings vs another car companies offering.

 

[RobDickinson]

More reliable? Checks Sandys bluecruise video. Sure.....

 

[voyager]

Are there any thoughts on the suitability of the type of car for self-driving?
Does a car need to ‘shape up’ in order to smarten up (become autonomous)?