How Cars Will Take Over the World
We see three stages for self-driving cars to become prevalent.
Stage 1 is where we are now. There are a handful of fully autonomous cars driving about, which are the research vehicles. These cars are limited to certain jurisdictions under certain rules until laws and regulations are sorted out to allow them to drive freely (with no human master) on the roads.
Stage 2, which will not likely happen until the early, perhaps mid-2020s, is where fully autonomous cars are being sold and there they share the road with human drivers. This is where things are interesting. We believe the fully autonomous cars with no steering wheel or pedals will be primarily purchased by delivery firms, fleets and taxi operators, while cars supporting fully autonomous modes with overrides for human control will still be sold to consumers at the high-end of the market. We think car ownership will start to fall considerably and it will make more economic sense for a consumer to subscribe to a mobility service or order a self-driving taxi when needed, rather than owning.
At some point, perhaps after 10 or 20 years, governments will choose a date at which point cars with steering wheels and pedals cannot be sold anymore and a later date at which it will become illegal for humans to drive except on designated tracks for hobbyists.
This starts the final stage where the road is only for self-driving cars. Car ownership shifts almost exclusively to fleets, companies or taxi services and mobility essentially becomes a utility for the consumer. Only the type of person (or company) that would like to own a private jet would own a private self-driving car.
We believe the world will be a much better place when this happens: most importantly, fatalities from car accidents will fall very significantly in the order of 95%+. Road capacity will increase significantly and traffic will become a thing of the past as cars can travel at high speed, close to each other, being fully and instantaneously aware of what all the surrounding cars are doing. Cars themselves will be used much more efficiently and be on the road for 95% of their lives, and the number of cars in existence will fall dramatically. Street parking will be a thing of the past. There will be more inner city green space. All cars will be electric, charging stations will be prevalent and global carbon dioxide output due to cars will be significantly curtailed.
I'm hopeful to see this in my lifetime (I'm in my early 40s) but I don't think it's likely.
The Market for Self-Driving Cars
I own a car which cost about $40,000, new. If I'm not working from home, it takes me approximately 45 minutes to get to work, 45 minutes back and maybe the car is used for additional half an hour during the day for going to the gym or running errands. That is an average of 2 hours out of 24 hours and the car sits idle for the other 22 hours. This is exceptionally inefficient for a product that cost me $40,000, which is depreciating at perhaps 20% per year, and has ongoing fixed costs such as insurance, road tax and maintenance.
Why do I tolerate this? Because it's convenient. I don't need to wait for a bus or taxi and overall journey times (in most cases) tend to be less than on public transport. It also provides me with privacy - I can make work calls. It is more comfortable and it is nice to have my own space.
Beyond this, I like driving. There is exhilaration in accelerating a car on an open road, in a car that I am fully in control of. It is like a large extension of me. And I'm not the only one - the wide popularity of The Grand Tour hosted by Clarkson on Amazon Prime demonstrates the popularity of power cars.
Car drivers are not interested in owning a taxi
Looking at a self-driving car, if we assume it will be sold as a high-end vehicle, we note that gross margins for the high-end manufacturers are typically in the mid-teen range versus the industry average of about 8-10%. Tesla (NASDAQ:TSLA), which can be seen as a premium car, shows a gross margin is in the low-20% range.
We argue that companies are not investing in self-driving vehicles just to command a higher sales premium. In fact, we don't believe that in the long term, consumers are willing to pay premium prices, or even buy self-driving cars at all. It might be a cool feature in the short term, but in the long term, there is no consumer market.
If my car was fully autonomous, then every time I need to go somewhere, I'm basically stepping into a taxi.
If a taxi service, say an Uber (Private:UBER), could pick me up in a clean and comfortable car within a few minutes of ordering the thing, with full privacy, without a driver, why would I need anything else? There is no reason why anyone would buy a taxi and leave it idle in the garage or a parking spot for 90% of the time.
Thus, the critical point is that the market for self-driving cars comes from the fleet operators and taxi operators, such as Uber, not from private individuals.
I own a car which is idle for 90% of the time. Taxi operators aim for close to zero idle time. In theory, my car could potentially service 10 other people, and perhaps more than that if I'm willing to ride-share. Though it is important to bear in mind that being on the road 24/7 will also mean higher servicing needs. However, at the same time, cars will be driven by their computer brains in a way which doesn't put stress on the machine.
This means, assuming I represent the average car user, in the self-driving world where cars are predominantly owned by operators and mobility services are leased, the number of cars needed to meet market needs can decrease down to about one-tenth of what it is now.
The Near-Death Experience for the Car Manufacturing Industry
The number of cars on the world's roads are estimated at around 1.2 billion and are forecasted to reach 2 billion by 2035. If by 2035, 100% of the global fleet is fully autonomous and mobility is sold as a service by operators, then we will only need 200 million cars to service all of us. This is basically a calamitous market decline for all the car makers and many, if not most, will die.
It is possible that gross margins on self-driving cars will be significantly higher than on regular cars (because of a large software component), which will take some of the stings of lower revenues. However, if there is significant over-manufacturing capacity, which will be the case at least for the first couple of years, these margins will be squeezed.
Thus, it is actually in the traditional car industry's interest to delay the inevitable for as long as possible. At the same time, it is in their interest to be the survivor and the survivor will make it big - the one or two selling 90% of the cars to the major fleet buyers. This means they need to invest or partner with someone investing in self-driving capabilities, and it's why they are all doing it.
The Complexity of the Problem of Autonomous Driving
If autonomous cars did not have to share the roads with human drivers and could organize themselves, the problem would be significantly easier to solve. 1,000 Apple (NASDAQ:AAPL) or Google (NASDAQ:GOOG) (NASDAQ:GOOGL) software engineers should have little problem in the solving of how to organize a network of cars where every car potentially knows the location of and plan of every other car on the road. Things like traffic lights, stop signs and other traffic organizing paraphernalia are done away with. The main issue is detecting, for example, a kid running into the road, something which is already done pretty well by semi-autonomous cars.
However, sharing the road with unpredictable humans that are not necessarily focused on the task of driving, communicate badly and make mistakes makes the problem exponentially more complex to solve. On top of that, humans will see autonomous cars as slow, law-abiding and easy prey and will take advantage of them.
It is in this in-between sharing stage where a lot of the self-driving R&D dollars are being invested. We will take a look at some of the main players investing these dollars and their motivations.
The 'quiet' Apple Way
In December 2016, Apple's Director of Product Integrity, Steve Kenner, sent a five-page letter to the National Highway Traffic Safety Administration (the NHTSA), commenting on the Federal Automated Vehicles Policy.
Until December, it had been an open secret that Apple was building a self-driving car. While Apple never made any public statements in this regard, quoting from CEO of Tesla, Elon Musk, "It's pretty hard to hide something if you hire over a thousand engineers to do it."
To quote from Apple's letter to the NHTSA:
"Apple uses machine-learning to make its products and services smarter, more intuitive, and more personal. The company is investing heavily in the study of machine learning and automation, and is excited about the potential of automated systems in many areas, including transportation."
This was basically the subtle Apple way of coming out of the self-driving closet, three years after launching the program. However, what is clear from the letter is Apple's interest is in the machine learning and automation - that is the software part of the self-driving ecosystem.
The background is that starting in 2014, Apple began working on a project. In February 2015, a first article appeared in the Wall Street Journal reporting that Apple was developing a self-driving car under the project name Titan, and was hiring up to 1,000 engineers to work including some very senior car industry executives with research experience. They were based in a secret location near its Cupertino headquarters. 1,000 Apple employees is a potential R&D investment in the couple of hundred million dollars per year, which means between 2014 and 2016, they may have spent around $1 billion on this project (which to be honest, is not that much money for Apple).
Apple also entered talks with and met with several car manufacturers with regard to a potential partnership including BMW (OTCPK:BMWYY), Daimler (DDAIY), Magna Steyr and even McLaren, but it appears nothing came of these talks.
In September 2015, it was reported that Apple had accelerated its efforts and the Titan project had received a "committed" label with a 2019 completion date for the project.
However, by the end of 2015, the project was hit by internal problems. In January 2016, Steve Zadesky, who was heading the program informed colleagues he was leaving the company. Later, in July, the Wall Street Journal reported that Bob Mansfield, a member of Apple's Special Projects team had taken over to lead Apple Car development and Bloomberg reported that Apple's car strategy shifted the project to focus more heavily on the underlying technology for autonomous vehicles rather than actually building an automobile towards autonomous driving. Apple laid off or transferred hundreds of employees who were working on the project following an internal 'reboot'.
Apple scales back on the self-driving car
Apple's letter to the NHTSA confirms Apple's new direction in developing the software for self-driving and would like to be the self-driving operating system. We believe Apple realized that becoming a car manufacturer is a whole different ball game.
While the iPhone does have a fairly deep supply chain, Apple maintains very tight control over its suppliers as well as the quality of the parts they manufacture. The supply chain for a car is exponentially more complex and it would be impossible for Apple to wield the same control over the suppliers it enjoys with its i-products. Additionally, Apple enjoys a gross margin at around 40% and the most it could likely hope for in an Apple car is half that, in the 20% range.
Thus, for now, Apple has rightly given up on building a fully blown self-driving car which would compete with the likes of Tesla. It is still focusing on software (it, of course, already sunk $1bn into the project). This makes more sense as software is a core competency that the company indeed has. However, the project as a whole is on probation and has until late 2017 to prove itself, otherwise, it will be shelved, which we don't think would be a bad thing for Apple.
The 'loud' Tesla Way
Tesla seems to be the furthest ahead in terms of commercialization of self-driving cars, or at least is one of the more vocal of the car manufacturers when it comes to self-driving.
Tesla started out fairly quietly in 2003 in Silicon Valley as a startup maker of high-performance electric cars. Its original vision was to prove electric cars can be as good, if not better than gasoline-powered ones. Its current stated mission is: "to accelerate the world's transition to sustainable energy", with sustainable transport being a subset of that goal. Part of what Tesla is doing is making the world a better place by building a greener car by using the (Nikola Tesla invention) AC induction motor rather than the combustion engine.
In 2006, it became a noisy company hosting a huge launch party in Santa Monica. Musk took over as the CEO in October 2008, and ever since the Company has not been shy about marketing, with massive launch parties and major PR announcements. Musk himself has a large messianic-like following with 6.5 million twitter followers.
In 2008, it shipped its first electric vehicle sports car, the Roadster. The company IPO'ed in 2010. In 2012, Tesla launched the Model S, a premium sedan and the Model X, a luxury cross-over SUV. In 2016, the Tesla Model 3, a smaller sedan was unveiled with deliveries expected to start at the end of 2017.
Tesla's self-driving foray
Tesla's foray into self-driving started in October 2014. Autopilot was part of a US$2,500 Tech Package option, and used Mobileye's (NYSE:MBLY) technology, initially offering semi-autonomous features such as auto-emergency braking and lane-keeping on marked highways, which feels like self-driving, but in reality are just advanced safety features.
In May 2016, a Tesla in autopilot mode had a fatal accident with a large white trailer and both human and machine failed to apply the brakes. Because Mobileye's system uses a monocular camera, it didn't see the white trailer against a brightly lit sky. Tesla blamed Mobileye, and Mobileye hit back saying its product is not a self-driving product, merely a safety feature and Tesla is improperly using it beyond its capabilities. The story concludes with Tesla ending its relationship with Mobileye in July 2016 and switching to an Nvidia hardware platform, adding a whole slew of cameras around the car as well as radar.
In October 2016, Tesla announced that all its vehicles 'have the hardware needed for full self-driving capability'. This does not mean the car actually self-drives, but it does mean that the cars will have the cameras and scanners, as well as the processing capability thanks to Nvidia, though the software (and regulation) still has a way to catch up.
What Musk actually cares about is saving the world
As part of its sustainable energy mission, in 2014, Tesla started building a Gigafactory to manufacture its batteries for storing the energy to power its cars with cell production expected to start this year.
The Tesla Gigafactory
By 2018, the Gigafactory is expected to reach full capacity. The goal is that the Gigafactory will produce batteries for significantly less cost using economies of scale, innovative manufacturing and reduction of waste. Cheaper lithium-ion batteries will remove a major hurdle for cars, or in fact, any machine using a combustion engine, to switch to cleaner grid or solar generated electricity.
While controversial from a financial standpoint, in November 2016, Tesla purchased SolarCity, a seller of solar power systems, which definitely does fit in with the company's mission statement.
While Tesla showed a profit for the second time in its history in the past quarter, we don't think Tesla, or more accurately Elon Musk, is all that interested in profits. Musk will invest what it takes to realize his long-term vision. He has no problem raising money when he needs it, and there are many investors more than willing to throw money his way at nose-bleed valuations.
Tesla/Musk sees a world where all cars are electric and driving themselves. If his dream becomes a reality, which we believe it will eventually, this will save millions of human lives and significantly reduce greenhouse gases. However, to justify the current market cap of $37 billion, at some point, an investor would need to see the possibility of a potential return on investment in the billions of dollars.
We don't see this point ever coming. We think Musk will prefer to invest any profits in furthering his vision of bettering the world. Tesla will spend a lot of money, will probably bring the world the first operational self-driving car, and we will thank them greatly for it, but Tesla will never really profit from it. We applaud the philanthropic nature of Tesla's investors.
The 'altruistic' Google Way
Alphabet, though let's call it Google, wants to organize the world's information and it does, in fact, do a fantastic job of it and makes a lot of money doing it. Google likes interesting and complicated software projects and a self-driving car is right up that street. In 2016, Google spun out its self-driving into a separate subsidiary, which is called Waymo.
Google's project is pure R&D and doesn't yet have any path to market. It is taking a slow and steady approach to absolutely full autonomy with no interest in launching a car with partial autonomy on the way. Google feels it would be doing a disservice to the world by releasing semi-autonomous as it leads to distracted drivers and potential accidents.
For Google, this is a technological challenge, and the company wants to do it right. In the past two years, like Apple, Google has also suffered some high-profile defections with many of the original project engineers leaving. According to a Reuter's article from about a year ago, the project has in the region of 170 employees which implies in the scale of probably a bit less than $100 million R&D investment per year. However, over seven years, this already starts adding up.
It started in 2009 under Sebastian Thrun, former director of the Stanford Artificial Intelligence Laboratory, whose team at Stanford created the robotic vehicle which won the 2005 DARPA Grand Challenge, where fully autonomous vehicles need to cross a stretch of desert with various obstacles in the fastest time.
By 2012, more than half a million kilometers had been driven and Google expanded from highway driving to more complex city streets. In 2014, it designed a new prototype vehicle with no steering wheel or pedals putting them on public roads in 2015.
The latest Google car
In 2015, Google achieved the first full self-driving road trip, taking a blind man for a drive in Austin Texas in a car without a steering wheel or pedals.
Google is investing R&D dollars here, though not on the scale that Apple was and taking a careful approach. This is one of its many projects it likes to take on which may, or may not, ever lead to a real business.
The 'safe' Mobileye Way
Mobileye's mission is to reduce vehicle injuries and fatalities by augmenting human driving with computer vision technologies to detect, warn and perhaps, prevent a potential collision before it happens. Mobileye has been the real winner in the current semi-autonomous period. It launched and sold its first chips for use by the car manufacturers in 2007 and has grown rapidly ever since.
The Mobileye EyeQ platform
The $50 million or so that it spends yearly in R&D is primarily focused on improving its existing semi-autonomous technology, which already generates significant revenues. Mobileye has obviously jumped onto the self-driving bandwagon and is making some investments on full autonomy built on its technology, but realizes it can't do it alone. Last year, it announced a partnership with BMW and Intel (NASDAQ:INTC), recently announcing a fleet of 40 autonomous cars to be on the road by the end of this year.
Mobileye's investment in self-driving cars is a natural evolution of its existing, highly profitable business, and it hopes to be the provider of computer vision processing chips for self-driving cars. However, it has found itself in the unfortunate position of competing directly with Nvidia for its slice of the fully autonomous space, and despite Mobileye's first mover advantage, Nvidia is seeing good traction.
The Nvidia 'Do What You Do Best' Way
Nvidia (NASDAQ:NVDA) is traditionally a graphics chip maker. These are processing units which offload complex 2D or 3D graphical processing from computer CPUs, and are primarily used for high-end gaming. At CES in 2016, Nvidia CEO Jen-Hsun Huang unveiled its Drive PX2 platform for processing the vision and radar data for self-driving cars. With expertise in graphics processing, it was not a major jump to for the company to launch a processor capable of processing image data and this chip was developed as part of Nvidia's ongoing R&D budget.
The Nvidia platform
In mid-2016, following the Tesla and Mobileye fight, it replaced Mobileye as the processor system inside Tesla's cars. The share price has reflected the success of Nvidia's self-driving processing platform, with the share price multiplying by over three times in one year.
The 'Don't get left behind' Car Manufacturers
The only ones that really have to invest in self-driving cars are the car manufacturers. They have a very simple motivation - the future will converge to self-driving cars and mobility will become a utility. If they are not there in some form, they will simply become irrelevant. Their interest is to remain ahead of the curve, but ensure it happens slowly.
All the major automotive manufacturers, including General Motors (NYSE:GM), Ford (NYSE:F), Mercedes Benz, Volkswagen (OTCPK:VLKAY), Audi (OTCPK:AUDVF), Nissan (OTCPK:NSANY), Toyota (NYSE:TM), BMW, and Volvo (OTCPK:VLVLY), have all been investing in and testing driverless car systems for the past few years. The general timetable being batted around seems to be the release to the public of self-driving cars by the early 2020s.
At this point, the market will start shrinking as individual people buy less and less cars. At the same time, the fleet/taxi operators buy in bulk, thousands of self-driving cars at a time. This gives the fleet operators a very strong bargaining position, combined with the fact that there will be significant over-capacity in car manufacturing. Despite the larger software component of cars (which will typically lead to improved gross margins), gross margins will be squeezed tremendously and many will cease to exist.
Some of the major car manufacturers are positioning themselves for this future crunch. While nobody knows what the space will look like exactly, they have been buying or partnering with tech startups involved in different parts of the mobility ecosystem. For example, GM acquired Cruise and partnered with Lyft (Private:LYFT). Ford acquired Chariot and computer vision startup, SAIPS.
The Start-up Players
There are indeed many startup companies in the space, and for a good reason. The space is very hot and valuations are extremely high. The better car manufacturers have money, and they need to be positioned for the future.
Some companies are being bought out almost immediately, in many cases, to access the expertise of the founders. For example, Uber, itself a startup, acquired a six-month-old self-driving trucking startup Otto and put Anthony Levandowski, its co-founder and one of the founders of Google's self-driving project, at the head of all its autonomous efforts.
The Winners and Losers
The losers will clearly be those that benefit from the existing paradigm: whereby people spend a lot of money on a major capital asset, a car, that is used for perhaps 10% of the time (stupid consumers like myself). This is most of the existing car automakers. The ones that don't evolve to sell self-driving cars will see their market slowly dry up.
Even if the manufacturers do manage to shift to self-driving, given the significantly improved efficiencies car sharing and mobility services will provide, there is only a need for a small percentage, perhaps 10%, in terms of quantities of cars circulating to service us all. That is a significant market shrink. Many of those that do evolve that don't have good enough relationships, or the ability to price their products and services low enough to fleet and taxi services to sell, will die anyway.
However, some of the high-end differentiated sports cars makers will be okay - people don't buy these cars for convenience, but because they want to own a car they will enjoy driving. In fact, some of these will likely become even more popular as car driving becomes more of a hobby than a necessity, and the only reason to own a car is for fun.
2016 McLaren 675LT
Tesla will probably be the first one out there with a working self-driving car. Whether it'll be one of the car manufacturers to survive the crunch will remain to be seen. However, either way, Tesla will probably move on to even bigger and greater things requiring oodles of money. If investors are willing to give it to Tesla, it'll probably still be around doing something interesting.
Google and Apple: They will obviously both be around. Self-driving for either of them, if they end up doing anything, will be a small and probably insignificant part of their business. Google will perhaps provide the Internet connectivity within the cars so it can feed the ads during transportation. Maybe Apple will provide the in-car entertainment. At the end of the day, there will be many more higher-margin phones in circulation than cars, and phones are replaced every two years or so.
Right now, it looks like Nvidia will become the processor system of choice for most of the car manufacturers. Thus, whoever wins will probably have Nvidia inside, and right now, we believe it is in a good position. Obviously, there could be plenty of startups out there that pop here and replace it, though it is clearly gaining traction and building a nice moat.
While Mobileye is the winner of semi-autonomous, to be a winner in fully autonomous, it needs to find a niche in that space and protect it. Perhaps this will be safety related or image processing. What it is doing with the REM (Road Experience Management) mapping system is very interesting, and maybe it will become the mapping standard for self-driving cars. In any case, we don't believe that Mobileye can compete with Nvidia on processing hardware. We would love to see it relevant in 5-10 years' time, we think it can do it and it will be interesting how.
We believe that the huge winners will be Uber, Lyft and other companies providing either fleet/taxi or more generally mobility as a service, let's call it MAAS (though I have discovered there is already a Wikipedia page for this term). Consumers will shift from owning cars to buying MAAS either by subscription or as a more expensive pay-as-you-go (basically a taxi).
For these companies, 75% of the costs are salaried employees - i.e. taxi or truck drivers and they only work for about a third of the day. Taking the cost of the driver out of income statement is very significant and then increasing capacity by 3x (that is each car providing 24/7 service), is phenomenal. We look to follow up this piece to see if we can identify some of the companies that sit it at this point.
There will be some winners and many more losers. The real winner of the gravitation to self-driving cars will be society as a whole and the consumer in particular.
Millions of lives that are cut short by car accidents every year will be saved. The transportation system will become significantly more efficient. Less road and car capacity will be needed and used much more efficiently. Consumers will save significant time as traffic and parking become a historical artifact. Greenhouse gases from the transport system will be reduced significantly. And finally, drinking and driving will become a thing of the past.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.