GM’s cruise unit has spend over $3 trillion on self-driving development, and just gave a first taste … [+]
Recent research at tech newsletter “The Information” calculates that the various companies have spent around $16 billion on self-driving car projects, calling it a “money pit.” (Story is behind paywall, but the free lede contains a chart with much of the information.) The bigger spenders were Waymo and GM/Cruise at over $3B each, and it didn’t even count for certain high value company acquisitions (for example, Intel spendng $16B on MobilEye with much of that value relating to their self-driving projects.)
The number seems astronomical, particularly in light of the expected recent pullback of enthusiasm from major auto OEMs for the rapid deployment of robocar technologies.
One still-enthusiastic OEM is GM. At their recent investor day, Dan Amman, CEO of the Cruise unit, described the eventual self-driving marketplace at around $8 trillion per year, with $5 trillion in rides, $2 trillion in cargo and $500B each for in-car experiences and data.
This $8T number is roughly in line with my own calculations for global ground transport. Of course, that requires that most ground transportation switch to being self-driven, which will happen eventually, but not for a few decades. Even so, numbers like that make spending $16B or more a drop in the bucket compared to the potential returns. Nonetheless, with numbers this huge and a future this variable, nothing is a slam dunk. Let’s examine what’s not in these numbers:
The first movers may not win
Everybody is spending this money because they hope the winner will collect the lion’s share of that $8T, and almost anything is worth that. As we’ve seen,though, in the high tech world, first movers often don’t win. Microsoft was first, and Apple one of the first, but Facebook, Google, Amazon and others were not first in their categories.
Everybody of jealous of Uber for having the world’s top brand in selling rides, since that’s the … [+]
What has everybody excited about robotaxis is that when you sell rides, instead of cars, you are selling the entire value chain of transportation. You sell the car, the fuel/energy, the maintenance, the insurance, the experience and everything else bundled into one price. You are the car makers, retailers and servicers and the oil companies and insurance companies and transport companies all in one. The company that controls the customer in that business (currently Uber or Lyft) controls where all that money goes. And during the period where good self-driving systems are rare, the company that controls the self-driving system has the power to own the customer. Yet, it may not remain that way forever.
In the early days of robotaxi deployment, there will be minimal competition. Each company will start deploying city by city, and they will tend to choose virgin cities with no competition over going head-to-head in a city that already has a player. It is possible that by the end of the competitive phase, there will be multiple strong self-driving stacks and it will be harder to control the world through that.
The price of car travel will go down, but the volume up
Automating car travel eventually makes the cost go down. In particular, the cost for the most common and basic of trips, the single person on a short urban trip. This is 80% of rides, and it can be served with small, low-cost vehicles very well, and it can also be served with shared rides during peak hours. Cheaper vehicles and sharing mean lower costs. In addition, today car ownership requires a large up-front cost which most of the world’s population can’t afford.
The good news for the auto industry is that if you make something cheaper and remove the up-front capital costs, vastly more people can afford it and will buy it, so while transportation gets cheaper, a lot more is sold. So that $8T may grow quite a lot.
The side benefits are immense
I calculate Americans spend about 50 billion hours driving cars each year. At the average wage in the USA (around $35/hour) that’s $1.7 trillion of productive time returned every year in the USA alone. A back-of-envelope calculation suggests a global value of this time or at least $5 trillion — Americans drive about 1/4 of the world’s miles but make an even larger share of its money. Even so, this matches the total price paid for car transportation around the world.
Robocars won’t eliminate all accidents, but they won’t be released until they are safer at driving than humans, and they will get even safer as time goes on. In 2014, the USA’s NHTSA calculated that accidents cost Americans $871B every year. Again, nobody has calculated the number for the world, but it’s going to be in the $3T to $4T range. Even cutting accidents in half produces astonishing benefits. (To be fair, driver assist technologies on regular cars are also going to reduce accidents, so much of this benefit is coming either way.)
More revolutions are yet to come
Some pioneers now feel that the ground was a mistake, and personal transport is going to move to the air. Sebastian Thrun, the most respected pioneer in the self-driving car space, has largely abandoned it and now preaches that the air is where the action will be, thanks to computerized electric vertical take-off aircraft. Air travel offers almost unlimited “lanes” of traffic in 3 dimensions, with trivial costs in infrastructure compared to the ground and is considerably faster.
This is just one of the things which might happen when you start trying to predict decades out. There are others which people are yet to think of which may change all these numbers in even more dramatic ways.
But it’s pretty sure that $16 billion is a drop in the bucket.
