Autonomous Driving – Pie in the sky pt. III

The pie has crashed back to earth.

• Another autonomous driving upstart is cutting back staff in order to preserve dwindling cash as revenues remain years away and current conditions will make it very difficult to raise anymore.
• This will exacerbate and accelerate the consolidation of the autonomous driving sector which I think has a bright future but still has far more players than it needs.
• Argo AI, into which both Ford and VW have poured money, is cutting 150 jobs and slowing hiring in what it calls “prudent adjustments to our business plan to best continue on a path for success.”
• In plain English, this means: autonomous driving is taking much longer than we promised and market conditions prevent us from raising more money so we will have to hold on as long as we can and hope we don’t go under.
• I am a long-term believer in autonomous driving because it solves a huge number of problems, but I have long expected that it will take much longer than expected to deliver.
• In 2017, RFM Research set a target of 2028 for autonomous driving to become a commercial reality (see here) which at the time was an outlier but is now well within the range of industry expectations.
• The problem was that targets for autonomous driving were set by executives who knew nothing about AI and assumed that the problems would soon be solved.
• The predominant method by which an autonomous driving system perceives its surroundings is deep learning which in reality, is nothing more than a sophisticated statistical pattern recognition system.
• This means that deep learning has no causal understanding of what it is doing and as such, is unable to think outside of the box.
• The net result is that whenever a situation arises that the machine has not been explicitly taught, it will fail to correctly interpret the situation which will cause the vehicle to make a mistake.
• The problem is that this occurs all the time on the open road which is why the speed of progress has been far slower than many had hoped.
• Furthermore, this slowness has allowed the laggards to catch up supporting my long-held view that when shopping for an autonomous driving solution, OEMs will have plenty of equally good offerings to choose from.
• This delay has been catastrophic because money was raised and valuations were set upon expectations which have now been shifted out by 8 years or more.
• Consequently, it is clear that the autonomous driving sector has been overvalued for years but may be reaching fair value given how the market has treated highly valued technology stocks with no revenue.
• 18 months I go I valued the global market for autonomous driving software at $44.3bn in 2021’s money (see here) using a 20-year DCF to 2028 and then discounting that back to 2021.
• If one adds up the mooted value of the sector even today, one will arrive at a value that far exceeds this number leading me to think that the sector remains substantially overvalued on paper.
• However, the public markets are telling a very different story as the market capitalisation of many of these sorts of companies is down from their listing prices by 80% or more.
• A haircut of this sort of size would most likely bring the paper valuation of the sector very broadly into line with the present value that I ascribed to it last year demonstrating that the public market is always correct in the long-term for the efficient attribution of fundamental value.
• Consequently, I think that the brutal consolidation of the sector is likely to continue as start-ups run out of money, are unable to raise more and are forcibly acquired by those with cash.
• The time is close for OEMs who currently do not have anything in-house and consider this activity to be strategic and differentiating to have a close look at what is on offer.
• In the sea of negativity, there is almost certainly some good value to be had.