Silicon Valley may have started out making silicon chips, but the Valley is protean. Chipmaking peaked in about 1985. After chips, we made disk drives. Then PC software, enterprise databases, and networks. Then dot coms and search engines. Then social, mobile, and SaaS. Now AI and data. The old stuff does not go away — there are still a few speciality chip fabs and plenty of fabless chip designers like Nvidia. We just pile on the newest stuff and keep going.
Up next: cars. Silicon Valley is making cars electric and autonomous. As the value of an automobile moves to its batteries, its network, maps, sensors, and AI, the car industry is moving to Silicon Valley. Thirty years ago, when a Business Week cover asked “Is Silicon Valley the Next Detroit?” it was predicting that traffic jams and high housing cost would lead to ruin. Nobody saw the car industry moving here.
Google now logs more than 10,000 miles weekly with self-driving cars — more than a million miles to date. Most of this is Mountain View (also Austin). Nobody else knows as much about navigating self-driving cars and building the underlying mapping technology as Google, although Uber is testing driverless car services. Assuming that Uber and its CEO can master the first rule of holes (when you are in one, stop digging), it is well-positioned to match cars and riders in hundreds of markets worldwide.
Electric car pioneer Tesla, based in the former NUMMI plant in Fremont, is widely expected to offer driving services that compete with Uber. Tesla is leading the move to electrify cars, and has caught a serious tailwind in the stock market. It’s market cap now exceeds GM’s, which is nuts, especially since the GM Bolt may turn out to be a breakout electric car.
Apple seems unlikely to sit this one out. They need a huge market to keep growing. On the other hand, their secret car project code named Titan turned out a bit too Titanic and sunk.
The current focus is on building an autonomous car that can drive down the street and not hit anything. This is the Google skill — and maps are critically important. Maps, he points out, have moved from discretionary accessory to vital necessity. Cars are also machine learning intensive — another Google strength, although not necessarily one they will control for much longer.
Autonomous cars need LIDAR, dedicated AI processors, powertrains, operating systems, simulators to test and refine software, HD mapping, software security, and Uber/Lyft type mobility platforms. These have all developed in the United States and more than a fair share of them in Silicon Valley.
As autonomous cars grow, they are likely to be deployed as a service by companies that can optimize a fleet of autonomous on-demand cars in a city on a real-time basis. This takes NetJet type optimization skills and seems well-suited to Uber or Google. It also radically reduces the number of cars (and parking spaces) the world will need. The industry will likely shrink as unit sales plummet faster than prices rise.
As cars go electric and autonomous, they will change the market in ways we can no more foresee than the guy in his buggy could anticipate the sexual revolution caused by horseless carriages. The car buyer is likely to change, which will change the car. Cars will become simpler, partly because electronic cars have about 10x fewer moving parts. At first they will still have a steering wheel. Eventually however, they may not have brake pedals or windshield wipers. And if you are summoning a fleet car, they will be purchased based not on Tesla-elegant styling, but by corporate fleet managers, much like PCs are today. They will not need or have recessed handles, leather upholstery, or other luxury touches.
High tech cars will be a large business, even if the industry shrinks. Indeed, that’s what attracts Apple and Google. There are simply not many industries left to revolutionize that can move the dial on companies with revenue measured in the hundreds of billions of dollars. Evans’ chart showing the place of the $1.2 trillion global car industry in the minds of technology CEOs is instructive.
As Evans notes however, this is today’s market — the one that looks to get smaller. “Some analysts are talking about unit sales halving over time (with growing demand from China and other newer markets offsetting new technology). Meanwhile, moving to electric can reduce the price of a car, or of course (Apple’s preferred option) expand margins.”
He notes that even if Apple goes after the premium car segment, as seems likely,
“the bubble on the chart above shows Mercedes-Benz, BMW, Audi and Lexus, which combined sell 5-6m cars a year for $220bn in revenues (and so averaging $40,000 per car). That’s where Tesla is aiming now, and where one might expect autonomous cars to arrive first. For comparison, iPhone revenue in the last 12 months was $146.5bn….To look at that another way, if Apple created a car business as big as BMW and Mercedes combined, that business would generate less profit than the iPhone.”
Cars shape everything around them, from families, to culture, social norms, and large adjacent industries. Half of global oil production today goes to gasoline, nearly all for cars. Removing that demand will hurt Russia, Saudi Arabia, Iran, and other lovelies (a fine reason for the DoD to subsidize our transition to electric vehicles). Cars kill more than a million people every year around the world, mostly due to human error. Fully automated cars would extend human life expectancy.
The car industry will also shrink because electric cars are simpler. No fuel injection, pistons, water pumps, or radiators. Evans: “Roughly half of US spending on car maintenance goes on things that are directly attributable to the internal combustion engine, and much of that spending will just go away.” Without mechanical or accident repairs, a large amount of economic activity disappears. 150,000 gas stations and associated convenience stores would have less demand and charging stations a lot more. Elon Musk wants us to charge our PowerWall either by day using surplus solar power or by night using off-peak power. If battery prices keep falling, this would begin to affect the economics of peak power generation construction by utilities — and a smarter grid would do the same.
Autonomous cars would not have to be totally safe to be a huge improvement over the status quo. Driver error causes about 90% of all accidents alcohol is involved with a third of them. Cars cause $240 billion of property damage and hospital/medical costs each year — which means that for every dollar we spend on a car, we spend fifty cents cleaning up the human cost of the car. As we eliminate accidents, Evans notes that we can start removing safety features: airbags, crumple zones, etc. that add to the cost and weight of a car.
What will happen to traffic? Automated driving should increase capacity, but with cars, increasing capacity creates more demand. Reducing congestion increases the number of drivers. Conversely, removing capacity can actually result in less congestion. Removing demand for off and on street parking lowers construction costs by about 18% and increases road capacity — assuming that everyone is going where they used to go, which did not happen when cars replaced horse-drawn buggies. As the cost of cars, fuel, and insurance drop and as driver costs go away, three-quarters of current operating costs disappear. More people will use cars as a service and not bother to own one. Busses may vanish in some places — who needs mass transit when door-to-door custom transit is cheap?
The big question is what happens to millions of affected workers? My quick estimate, using BLS data, suggests that perhaps four million jobs could be affected.
|Total||~4.4 billion||$156 Billion|
|Heavy Tractor Trailer Truck Drivers||1,678,200||$42,000||A
|Light Truck or Delivery Drivers||826,510||$34,080||B|
|School Bus Drivers||505,600||$30,580||C|
|Rental Car Agencies||447,050||$28,210||D|
|Auto Body Repair||256,620||$43,000||E|
|Taxi Drivers and Chauffeurs||180,960||$26,070||F|
|Transit and Intercity Bus Drivers||168,620||$40,160||G|
|Uber and other on demand drivers||160,000||$30,000||H|
|Parking lot attendants||144,150||$22,520||I|
Four million sounds like a lot of jobs. Is it? How long does it take the US economy to generate four million jobs? From one perspective, the answer is less than a month, since the US economy generates about five million new jobs each month. Of course, it also destroys almost that many — so self driving cars represent perhaps a month of typical job destruction.
Of course self-driving cars will arrive over a period of years and decades. Further, not every job on this chart would be affected equally. Local delivery requires loading and unloading, so if the driver doesn’t do it, someone or some robot will. Meter maids may not all disappear, and one could argue that if my main car is an on demand service that rental car services will go up, not down. Many of these jobs are disappearing anyway. The long haul trucking industry claims to be short 50,000 drivers as it is. These jobs are not especially fun, healthy, or good for those who do them. Will truck drivers face mass layoffs? Probably not — this is the sort of change that can be managed with attrition, especially with a workforce that is already quitting faster than companies can hire.
As with cars we drove, the biggest changes will be the second and third order effects. More fortunes grew out of real estate values changed by cars than from selling cars. Cars remake cities and self driving cars will remake them as well. Commuting will change. Drinking, texting, watching movies, and working while driving will change.
Ben Evans also reminds us that self-driving cars are not just big computers, they are big cameras.
“Pretty much every vision of automatic cars involves them using HD, 360 degree computer vision. That means that every AV will be watching everything that goes on around it – even the things that are not related to driving. An autonomous car is a moving panopticon. They might not be saving and uploading every part of that data. But they could be.”
Soon enough, the police will be interrogating cars who were in the area as to whether it saw something suspicious.
Sources for the table on jobs affected by self-driving cars:
- A: BLS OES 53302 at http://www.bls.gov/oes/current/oes533032.htm
- B: BLS OES 533033 at http://www.bls.gov/oes/current/oes533033.htm
- C: BLS OES 533022 at http://www.bls.gov/oes/current/oes533022.htm
- D: BLS OES 412021 at http://www.bls.gov/oes/current/oes412021.htm
- E: BLS NAICS at http://www.bls.gov/oes/current/naics5_811120.htm based on estimates of 90 percent fewer crashes with self-driving cars.
- F: BLS OES 533041 at http://www.bls.gov/oes/current/oes533041.htm
- G:BLS OES 533021 at http://www.bls.gov/oes/current/oes533021.htm
- H: Jobs from http://www.wsj.com/articles/uber-touts-its-employment-opportunities-1422229862. Income estimated to account for large share of part-time drivers.
- I: BLS OES 536021 at http://www.bls.gov/oes/current/oes536021.htm
- J: BLS OES 333041 at http://www.bls.gov/oes/current/oes333041.htm
Note: parts of this article appeared in a different form in 2015