Autonomous dreams

It’s not deliberate that I link the blog to Tesla for the second week running, but there’s always a lot going on when Elon Musk doesn’t apply a filter to any of his comments.  Last week, Tesla announced their second quarter results with the key highlights being a 10% core automotive operating margin despite the succession of price cuts, volume ambition this year of 1.8 million (double those achieved in 2021!), and a promised start of production date for the Cybertruck in the final quarter of this year.  However, Musk also talked about the company’s commitment to AI (Artificial Intelligence), partially for application in the company’s Full Self Driving (FSD) capability.

FSD has attracted a lot of criticism – from customers who have paid for it on their current cars but not yet received the promised functionality, and from some regulators and lobbyists who point to a succession of incidents where FSD appears to have malfunctioned resulting in death and serious injury.  Musk’s response on the first point has been to offer a “one time amnesty” to customers who had paid the fee on their current Tesla, to transfer that entitlement to a new Tesla as long as they take delivery by the end of September – effectively a lever to allow Tesla to hit their quarterly sales volumes.  As things stand, when you “buy” FSD, you are actually subscribing to it for the period of your ownership or lease – when you change cars, the functionality is removed from the car, and you have to pay again on your next Tesla.  The cost is not trivial – currently US$15,000, up from US$5,000 when it was first launched, and an estimated 400,000 customers have bought the feature so far.  (Most of these comments relate to the US – practices and costs will vary between markets).  On the second point, Musk views things from what I would see as an actuarial perspective – that incidents related to FSD are fewer than those that would occur if a human was in charge.  Not sure that this assertion has been factually proven, but it seems that he basically accepts that it is OK for his technology to kill some customers as long as that number is fewer than would kill themselves through their own incompetence.  That argument would not get past investment approval in a company like Mercedes or Volvo who would be targeting zero failures.

In the results call, Musk then said that “the short term variances in gross margin and profitability really are minor relative to the long-term picture.  Autonomy will make all of these numbers look silly.”  His claims are based on the income from selling FSD functionality, and the potential for owners to earn an income from their Teslas by them operating as autonomous robotaxis when the owner does not need them.  The company is also – they claim – in discussion with another OEM about licensing their FSD software, so the revenue promise is not restricted to cars bearing the Tesla brand.  Musk said in June that solving full self-driving “is really the difference between Tesla being worth a lot of money or worth basically zero.”  When the company is currently valued on the stock market at more than ten times that of Ford, GM or VW despite the uncertainty, you wonder what “a lot of money” means to billionaire Musk…

When other car manufacturers and ventures such as Waymo have scaled back their autonomous car ambitions in terms of both capability and timeframe, you have to ask what Musk knows that they don’t to justify an ambition of being ready in months rather than years (or even decades).  I guess that part of this relates to the fact that he has been here before.  Little over ten years ago, few people believed that he would succeed with establishing a viable electric car company.  Fifteen years ago, it seemed highly unlikely that SpaceX would overcome early failures and reach the point where it can reliably offer orbital launch services to blue chip clients including NASA themselves.  People are just slightly suspicious that he might have an ace up his sleeve, and repeat this trick for a third time.

I’m actually a fan of autonomous driving capability.  As those who know me will recall, I had the Autopilot feature on my Tesla and valued it highly.  Recognising its limitations, I still found that it made for much more relaxed motorway journeys, particularly in stop-start traffic conditions – the sort of traffic conditions where even the most enthusiastic driver would find it hard to argue that they derived any pleasure from the experience.  In my view, this is where the real value is in autonomous capabilities – removing the stress from everyday driving situations, and largely eliminating low speed accidents that result from boredom and distraction.  How many people are going to be happy to send their car off for a night shift as a robotaxi, then have to deal with the consequences in the morning?  How many people have an end-to-end journey that does not include some chaotic environments such as busy car parks and complex junctions where cars, bicycles, pedestrians and even trams come together?  The dream of autonomous driving might turn into a nightmare as you sit marooned in the middle of the junction with your car yielding to every other road user.  (And I might have paid Elon $15,000 for the privilege of doing this for three years…)