Waymo – The Supplier

Waymo will enable Robotaxis, not operate them.

• Waymo’s application for a licence to operate in New York City is more about data gathering and working out how to operate in one of the most complex road environments in the world, and less about an imminent strategy to become a robotaxi operator.
• Waymo has applied for a permit to operate its autonomous vehicles in New York City which will involve a safety driver sitting in the driving seat.
• This is a legal requirement as opposed to a technical one, as Waymo operates with remote safety drives in the other regions where it is present (San Francisco, Bay Area, LA and Phoenix).
• This is unlikely to change until other markets are fully up and running and the regulators of New York realise what they are missing out on.
• This is being seen by Lyft and Uber as a statement of intent to enter the Robotaxi market but I think that this is very unlikely.
• This is because the Robotaxi market, when it comes, is not going to be a high-growth, high-margin market where everyone does well as Tesla seems to think.
• Instead, it will be high growth for a while, but it will also be a bloodbath of brutal competition because there will be very little differentiation between the different contenders.
• There are several different players in the autonomous driving space, which are now pretty much as good as each other.
• Surprisingly, I do not think that Tesla is one of them and if Tesla was to produce performance figures rather than anecdotes, I suspect it would be rated as a fast follower rather than a leader.
• This is because Tesla refuses to use Lidar, which in theory should not be needed, but because the machine vision system keeps making mistakes, it is needed for error correction.
• This is why all of the leaders in autonomous driving use Lidar, and those that do not remain somewhat behind.
• Everyone is now pretty much the same because the progress in improving performance has slowed to a crawl which gave those who were somewhat behind an opportunity to catch up.
• As a result of everyone being roughly the same, when commercial robotaxi services are launched, there will be very little to separate the different services from one another.
• Hence, the economics are unlikely to be $1 per mile with a cost of $0.25 per mile as Tesla claims, but more like $0.33 per mile with a cost $0.25 per mile.
• This means that revenue forecasts for robotaxis are overstated by a factor of three and gross profits by a factor of nine.
• From a valuation perspective, this means that using Tesla’s numbers, a valuation of its robotaxi business is massively overstated and given that humanoid robots are unlikely to be a viable market for many years, I continue to find Tesla’s share price very challenging.
• From Waymo’s perspective, this reality means that there is very little sense in it entering the market for robotaxis, especially as it does not currently make cars or provide transportation services.
• Instead, I think it will focus on providing its driving solution to companies that wish to make self-driving cars or provide robotaxi services.
• This will be a much smaller opportunity in terms of total revenues, but given how cutthroat I expect the robotaxi market will become, I suspect that most of the profit will be made supplying the technology rather than using it.  
• Waymo’s current problem is cost, as with all the sensors that it uses to detect the environment and the compute to process it, the setup remains expensive.
• Wider adoption and deployment to private vehicles will require a cheaper implementation, which means reducing sensors, which in turn will require the AI that makes sense of the data to get better.
• This is going to take some time, and in the meantime, we are going to continue to see cautious roll-out of advanced testing in several markets, as I do not agree with Nvidia’s view that a deluge of synthetic data will solve the problem.
• Consequently, I am sticking to my long-held target of 2028 as the time when robotaxis may start to appear in more volume, but this remains a sector of the market I would not touch.
• Instead, the technology suppliers look much more interesting but at the moment, only PonyAI and WeRide, which are both Chinese and consequently raise all of the usual concerns, are available as direct investments.
• If I were to go anywhere, it would probably be Google, which is inexpensive, is fighting the AI fight much better than a year ago and is still showing healthy growth.