Tesla FSD – The Mannequin.

A dummy will be launched.

• The idea that Full Service Driving (FSD) will be a commercial reality in 2025 remains a fantasy but Mr Musk will strengthen the resolve of his fans with the launch of a vehicle that will look like FSD but lack the intelligence to offer FSD in any commercial capacity.
• In its most recent earnings report, Tesla officially pushed back the date of the FSD vehicle launch to October to give the design team more time to perfect the vehicle.
• It is important to remember that this delay is not to perfect FSD but the physical design which I think is a sign that FSD itself is not close to being commercially ready and remains way behind its peers.
• The evidence for this is everywhere and begins with the fact that Waymo, Cruise and many Chinese players already have commercial or semi-commercial vehicles on the road while Tesla has none.
• Furthermore, recent tests of FSD still display basic errors that indicate that FSD remains unable to deal with anything that falls outside of events for which it has been explicitly trained (see here).
• This problem exists for all systems that use machine learning to power vision systems but everyone else also uses Lidar which makes it much easier to reduce the rate of errors.
• Mr Musk is correct when says that lidar is a crutch and that a machine should be able to drive a vehicle using just cameras (like humans can) but he is assuming that there is a functioning brain on the end of those cameras.
• This is where he goes wrong and by not using lidar, he is making it much more difficult for his vehicles to correctly perceive the environment which is why I think he remains significantly behind everyone else.
• Hence, I suspect that the October event will reveal a design for a robotaxis and a rehash of the business model that in my opinion remains fundamentally broken.
• Tesla correctly (in my opinion) expects that the cost to provide a robotaxi will be around $0.30 per mile but incorrectly thinks that the price will be $1 per mile.
• This is how it thinks that it will be hugely profitable in robotaxis and is where its long-term growth expectations come from.
• I think that many of the other players such as Waymo, Cruise, Nuro, Pony, Baidu, Mobileye, Avride and a number of other Chinese players are better at autonomous driving than Tesla.
• Consequently, I rate Tesla as second rate at autonomous driving in the best instance meaning that it is not going to be the first to market, which is why its economic assumptions are completely wrong.
• Hence Tesla will not have the market to itself when it finally arrives meaning that robotaxis, like food delivery and ride-hailing, will be a commoditised hotbed of brutal competition.
• This will lead to the $1 per mile price rapidly collapsing to something more like $0.4 per mile or less meaning that this will descend into a fight to the death where only the biggest and most efficient survive.
• With a price of just 40% of that expected and gross margins of 20% rather than 70%, it is easy to see how Tesla’s expectations for robotaxis are one or more orders of magnitude too high.
• Consequently, I think that the market’s view of Tesla’s robotaxi business is far too optimistic and that instead of being the panacea that cures all of Tesla’s ills, it could be a near-fatal disease.
• Hence, I still think that Tesla’s valuation is far above where it should be given the business that it has and its outlook and I continue to avoid the shares.
• There are much better places to look for investing in electrification and autonomous driving like nuclear power and overly sold lidar companies where I have a position in Ouster.