Get email delivery of the Cadence blog featured here
In New York, there is an occasion four times a year known as the "triple witching hour." JK Rowling might have come up with the phrase, but, in fact, it is the last hour of trading on the third Friday of the last month of each quarter. On those days, stock market index futures, stock market index options, and stock options all expire.
The automotive industry is facing its own triple witching hour, but one that happens more like once a century. But there are three things potentially expiring at the same time, although not in a neat little hour like on Wall Street.
It is increasingly looking like the three things that are expiring are:
In some ways, electric traction for vehicles is an orthogonal dimension, nothing to do with the other two things. But it is very disruptive. Existing car companies have a lot of expertise in building internal combustion engines that have high mileage, good acceleration, and low emissions. Obviously there are some other assets existing car companies have, such as supplier relationships, building suspension systems, comfortable cabins, dealerships, and so on. But a huge barrier to entry for a new car company has always been how to get engines. As if to emphasize the point, there are no engine manufacturers (except for things like Formula 1), there are just car companies. Some low-volume companies buy engines, but not from pure-play engine makers, from other car companies. The equivalent of the fabless/foundry split in semiconductor never happened in automotive.
But electric power is a comparatively simple technology from a mechanical point of view (maybe only a single moving part) and the differentiation is in controlling the power and the mileage-per-charge, which is basically all software. Marc Andreessen wrote in a famous WSJ article that Software Is Eating the World. I've heard him explain what he meant. Obviously not that cars are just going to become apps on our phones. Rather, the winners in the future automotive markets will be the car companies with the best software.
With comparatively simple electric traction, mastering internal combustion engine technology will become as much a liability as an asset. This opens the door to new entrants. It is a door that Tesla has already walked through, and other companies such as Waymo (Google) or Apple seem to be getting ready to. Maybe someone will arrive who has yet to even poke their head over the parapet. The existing car companies all have electric programs of some sort, with varying levels of success and commitment. General Motors, perhaps surprisingly, are one of the leaders with the Chevy Bolt (called the Amera E in Europe), which is estimated to be shipping at 20,000-30,000 vehicles per year, but being first to market is not always the route to eventual success (remember Sidecar, anyone?).
Of course, the death of people driving their own cars is just another way of talking about autonomous driving. There remains a lot of discussion about just how fast this will arrive, but I think there is general agreement that advances in the technologies required for autonomous driving, especially machine vision, are progressing much faster than anticipated. One of the first posts that I wrote when Breakfast Bytes started was Ten Years Ago Self-Driving Cars Couldn't Go Ten Miles and, although the number is up to 13 years, it is still amazing progress in a very short period of time. I have seen people argue, and I have to agree from what I have observed, that there has been more progress made in machine vision in the last three years than in the previous...well, since people first tried attaching crude image sensors to computers.
Many observers will give you their estimate of when the technology will be good enough for autonomous cars to be a reality, at least except in the most extreme conditions such as snow on the road. But they often go further and give their estimate of when it will become illegal to drive your own car, either because the law makes it so or because no insurance company will touch such a risk any more. In some ways, it is surprising we let teenagers control two-ton vehicles. I think the US has the lowest age, letting teenagers as young as 15½ drive—and remember, we don't consider them competent to make decisions about their own bodies at that age (at least the law doesn't, teenagers often have their own ideas). But by 17 or 18, almost all countries let teenagers drive.
Currently, it takes about a decade for most of the fleet on the road to be replaced, meaning that most of the cars that will be on the road for the next 10 years are already on the road now, and very few of them are designed to be upgradable with future technology to be autonomous. So non-autonomous vehicles are not going away overnight. However, this is the triple witching hour, and if fewer and fewer people own their own cars, then the historical number of vehicles could drop and old vehicles be phased out much faster.
Young people are often setting their lives up so that they don't require a car. They just use Uber or Lyft, along with walking, cycling, or taking the bus. One of the attractions of living in a big city, not just New York any more, is that you don't need to own a car. My kids are a little bit older than that and wanted to drive as soon as they were legally allowed to, and wanted their own car. In fact, my daughter was so insistent on learning to drive that I taught her the basics on the high school parking lot in my stick-shift Miata when she was 14. But I keep reading how teenagers today are more like teenagers in New York City used to be 20 years ago: they assumed they'd learn to drive one day, but it wasn't a priority.
Of course, most of us still own our own cars. Partially, I suspect, that is out of habit, partially because we chose to live and work in a way that assumed we owned our own car, and partially because it really might be the most cost-effective way to get around. It is certainly more convenient, but I'm not sure I am saving any money by not getting Uber to work every day. I recently heard a podcast with James Altucher, who has given up his apartment and just uses Airbnb even in New York City, where he nominally lives. Doing the equivalent for your car seems a lot less extreme.
These three trends are arriving over an extended period. So maybe it is a triple witching decade. But they are a huge threat to the companies that make up the automobile industry today. First, the biggest factor that makes an existing car company competitive will go away, and gradually cars will become electric. Second, a key technology for the future is one that no existing car company already has as a core competence, namely software.
But the third one is the big one, enabled by the other two. The car companies today have a huge benefit in selling us all very expensive vehicles that we use with a duty cycle of roughly 2%. Our cars spend most of the time in our garages, or in the parking lot at work. If owning a car becomes more like owning a plane (or a horse), something for the very rich or as a hobby, then the size of the automotive industry could be 30% or even less of what it is today. In every country, cars are a big market (high price and lots of vehicles) and so governments have tended to want one of their own, and are very reluctant to see existing plants close down even if the company is insolvent. So there seems to be a lot of overcapacity, especially in Europe (with lots of countries and brands: Peugot, Citroën, and Renault in just France, for example).
If the industry shrinks significantly, even a very conservative 25%, then many of the weaker companies will not make it. It won't be that there are the same number of companies and plants, just running at lower capacity. I think it will be more like what has happened in automotive early in the 20th century. A lot of consolidation and, as the economic volume for a car plant climbed, a lot of companies unable to afford a competitive manufacturing facility. There is room at the high end for some specialized low-volume manufacturers like Ferrari or Bentley, but the mainstream has to be cost effective. It is still a spectrum, you might still pay twice as much for a Tesla or BMW as a similar(ish) Toyota or Ford, but not 10 times as much.
It is going to be interesting to watch.
It is automotive blog posts all week, with two double headers: first, two posts about CactusNet, based on Michelle Mao's presentation from Autosens in Detroit. Then two posts on Robert Schweiger's presentation at CDNLive Munich that kicked off the automotive track.