Robotaxis – Nasty endgame

The industry is inching towards a bloodbath.   

• All of the larger players who intend to offer robotaxis services are inching towards commerciality but when they finally get there, they are likely to find a bloodbath of brutal competition from which only a few survive and far less profit than expected is earned.
• Ever since Elon Musk fired the starting gun on the autonomous vehicle race back in 2016 the mad rush has slowed to a trickle as it turns out that autonomous driving is much harder than anyone thought to get right.
• Autonomous driving systems use deep learning to make sense of the signals from cameras, lidars and so on and, because of this, these systems have exactly the same problem as larger generative AI systems.
• This problem is that they have no causal understanding of what it is that they are doing and, as such, are incapable of dealing with anything that they have not been expressly taught.
• In practice, this means that if anything unusual happens or the environment changes, the machines are unable to deal with it and make perception mistakes that even the dumbest humans would never make.
• It is this problem that has meant that the promises of autonomous vehicles have not been met and interest and investment in the sector have collapsed.
• A lot of small start-ups have left the market but those with big backers are still in business but their progress towards providing a commercial, revenue-generating service is painfully slow.
• Furthermore, every time something goes wrong, and people get hurt or put in danger, the progress slows down even more.
• The whole idea of robotaxis is that by removing the human from the product and by shifting from petrol to electric, the service becomes much cheaper to provide.
• On average, ride-hailing costs about $2.50 per mile to provide but this could be reduced to something closer to $0.40 per mile.
• Mr Musk’s contention has always been that he would be able to charge $1 per mile and thereby make 60% gross margin which has been used to underpin Tesla’s valuation.
• In my opinion, this required Tesla to be the only provider of a robotaxis service, but it is clear that by the time Tesla gets there, the market will already be occupied.
• This is because Tesla is not a leader in autonomous driving but sits in the middle of the pack with a bunch of other players.
• The slowdown in the development of autonomous driving has also meant that those that were previously laggards (including Tesla) have caught up with the leaders and there is very little to differentiate between them.
• Progress is painfully slow but Cruise is back, Waymo has expanded in Los Angeles, Zooks is now offering rides in Foster City, Baidu has over 600 vehicles in China, Pony.ai is testing in designated zones in 4 Chinese cities while WeRide is testing a variety of services in 26 cities worldwide.
• Mobileye is also working on a robotaxis offering as are a number of others outside China and the USA.
• This means that when the technology is finally ready, there will not be one player in each market but several, with very little to tell the difference between them.
• Consequently, the realised price will not be $1 per mile but will quickly erode to something closer to $0.48 per mile or even less.
• Gross margins of 20% will be reduced to mid-single digits after operating expenditure is taken into account, meaning that robotaxis will not provide high profitability but instead be commodities.
• Hence, it is likely to be a bloodbath of brutal competition where only the largest and most efficient providers survive.
• This is not a particularly good outcome for an industry that has spent billions of dollars in the hope of high returns, which is why I would not want to have financial exposure to any of these players and would instead look to invest in companies that benefit from growth in greater autonomy.
• This is why I own Ouster (digital lidar) which has had a very rough 2 years but the company has prevailed and is in a good position to grow its way to profitability without having to sell a single lidar to a car company or raise any more money.