Tesla is rolling out FSD (Full Self Driving) version 13 this weekend to Tesla employees. This includes a fleet of Tesla cars being used to give Waymo-Uber like ridehailing to Tesla employees.
I make the case that FSD Version 13 and the systems that follow next year will give Tesla a 5-10 year lead. The car company and Waymo competitors in China are currently buying Nvidia Orin and Thor driving chips and software. The Nvidia driving chips are just the base system and the companies still have to train them to create their driving automation AI.
IF Tesla has a 5-10 year lead then the obvious choice to catch up with a system that is 600 times better than the Ford, Xpeng, Huawei, BYD or other driving systems is to license FSD from Tesla.
FSD Hardware AI4 Chip is Needed for Robotaxi
The FSD Hardware 4 computer uses a custom System on a Chip (SoC) called “FSD Computer”. Tesla has been trying to use the Hardware 3 chip for robotaxi capability but will likely have to replace it. Tesla said they would replace it for free.
Catching up to Where Tesla Was at the End of 2023
Tesla had the AI4 chips installed in 5 million cars at the end of 2023.
This cost about $10 billion for the chip and camera hardware alone.
This was $200 billion worth of cars which customers paid and Tesla made profits.
All other car companies sell electric cars for a loss.
Spent years developing up to FSD version 11.
Tesla had to convert to end to end neural net in 2023.
Tesla had to unify city driving and highway.
Tesla still needed 2 billion miles of driving data in 2024.
Tesla created superior testing and simulation systems.
Tesla automated labeling so no human labelers were needed.
They mostly automatically identify all of the learning cases.
They pull gigabytes of data each car from every day. Hundreds of petabytes.
They have proper scoring of the quality and safety of human drivers. The AI then copies the best human drivers.
The choice is to Spend $10+ Billion to try to catch up to Tesla on a multi-year project or license an FSD the will have solved robotaxi.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
HW3 in about 4.5 million Tesla’s. HW4 about 1.5 million, since being released early 2023.
Uptake of FSD is 10 to 12%.
So the cost to Tesla to upgrade HW3 to HW4 for vehicles with HW3 that purchased FSD is about $1 billion, not $10 billion.
Anyone with HW3 that doesn’t have FSD, will likely also have to pay for the hardware upgrade.
Please stop.
So is there any reason to get Self Drive if it is not (yet) good enough for me to lie back and nap until the car gets to where I want to go?
Depends on the person. We just bought a used Tesla with FSD 2 months ago. I find FSD beneficial, and use it nearly 100% of the time. On freeway definitely feel it is a better driver than I am, and I’ve been in one accident in 40 years of driving that was the other person hitting my car. In town, it’s likely a bit safer, but I use it as much to see how it will work in certain situations. After a couple of weeks, I feel comfortable taking over when necessary and when the system requires an intervention. But I could see where some people would find this difficult or stressful.
Example: My wife doesn’t use it at all. But she also never used cruise control and any other driving features in any car we’ve owned. I find such things convenient and make driving more relaxing.
no.
I think someday, it will be valuable, but not today. Maybe late next year, but even then, only for their newest cars.
It is already valuable. Especially on longer freeway rides. As soon as the Cybercab production starts, you know, the software is ready.
One wonders how easy or political it will be to license FSD in a state or region. Interesting if roll-out in some areas is held up for years or decades based on the prevailing regulations. FAA and SpaceX is one thing, but autonomous cars driving throughout some small towns, upstate? Could be the temporary hold-back that counter balances any lead Tesla ever had on competitors, in such a world where FSD5 and robotTaxi become significant (not my prediction this decade). I look forward to see if the technology, demand, and resulting ‘driving environment’ ever get good enough.