The wave of fervor over self-driving cars is putting Uber in a precarious position. At first blush, it is easy to believe that fully autonomous vehicles will be the step that propels a valuable company to unfathomable heights. But a deeper look reveals that self-driving technology poses a strategic problem for the company.
If unsupervised, fully autonomous cars fill the road too quickly over the next several years, Uber will be far more vulnerable to competitors like Google and Apple—who have both lower capital costs and the ability to acquire customers cheaply.
In this respect, Uber’s move to be the first to put self-driving vehicles in a commercial capacity on the road in Pittsburgh feels like a necessary but scary acceleration of the self-driving game by the company. Uber might competitively need to do it, because if it doesn’t, someone else will. But if Uber pours too much fuel on the self-driving fire, it will put itself at a strategic disadvantage.
Supervised Self-Driving Tech: A Boon for Uber
A world in which cars mostly drive themselves, but licensed drivers are still required to be at the wheel, is probably the ideal outcome for Uber in the next decade.
First, a regulatory requirement to have licensed drivers will allow Uber to continue to leverage perhaps its greatest competitive advantage at this point. That is its robust driver network, and the brand’s ability to acquire new drivers cheaply and efficiently. This advantage will be hard for any pure software or hardware company—ranging from Google and Apple to the Detroit players—to overcome.
Second, supervised self-driving technology will likely reduce the cost of driving and expand the pool of drivers Uber can draw from. Self-driving tech, just like mapping tech, makes the job of driving around the city easier and more accessible, which should expand the attractiveness of the job and the labor pool. Further, if supervised self-driving technology helps avoid accidents, insurance costs for drivers will drop, generating a savings which can be passed onto customers and further expand the market or taken as profit. As a market maker for transportation transactions, a bigger safer labor pool, and lower costs per mile, this technology would mean more customers, more scale and access to lower-cost capital.
Finally, supervised self-driving technology suits Uber’s strategy for financing the cars—the hardware for its network. People forget that Uber is a financial platform as much as it is a network for drivers and passengers. Rather than being asset-heavy, Uber leverages the credit of its drivers—and the service companies that have popped up around the platform—to purchase, service and operate its virtual fleet. Supervised self-driving doesn’t break this dynamic because in this world drivers are still in the loop and drivers still probably own their own cars. Uber gets to keep leveraging the buying and financing power of their driver network rather than having to take on the capital expense of buying the cars itself. (Lyft's co-founder John Zimmer yesterday published this post outlining how self-driving cars would benefit Lyft.)
Truly Autonomous Cars: Threat to Uber
Autonomy will open up the competitive floodgates. Once you have truly autonomous vehicles without drivers, providing transportation for people or goods is going to be a game of cost of capital and customer acquisition cost. The value of having a driver network will drop to zero rapidly, eroding one of Uber’s core competencies. From a consumer standpoint, it may still have a lead in initial customers. But to enter the market, Google, Apple or Amazon need only to build and deploy the hardware needed in a city and send out a bunch of push notifications announcing their new service. If you can afford to buy the resources you need, market entry gets way, way easier.
At the same time, it is extremely unlikely that any player will have a real technological advantage over anyone else in self-driving tech. In a completely idealized model of the world, when there is a very high penetration of self-driving cars, the mesh-network approach where cars can talk to each other should provide some sort of network advantage where there is return on scale. Shorter term, there will be some mapping advantages to having more cars already deployed. Sensors on cars should help a self-driving fleet identify road characteristics, traffic patterns and potholes faster.
But, in reality, full autonomy will have to roll out starting at zero percent of an enormous universe of un-networked cars on the road and build up. This means that all the tech will need to be self-contained to each car, and the network effects will be weak.
The software that is going into self-driving cars will be hard, but hardly the exclusive intellectual property of any one company. This is, first, because tons of smart people are focused on it. Some people are smarter and faster than others, but everyone will reverse-engineer each other into an even playing field.
If you don’t believe that argument, consider that it is extremely unlikely that the government will let a single company monopolize fully autonomous driving technology. Anyone who produces it will almost certainly have to share it with others to get regulatory approval to use it on the roads. It is too powerful an advantage and too tightly regulated a space for any one big player to get away with building a huge lead.
So, the world I see is that once true autonomy happens, whoever has the lowest cost of capital for building, deploying and managing cars as a capital asset on their balance sheet and has low customer acquisition costs is going to win. What they win will be a real question, because it is likely that competition will be so fierce that these will not be high-margin businesses. But any way you slice it, it is going to take Uber at least a decade to grow to a scale—and build up sufficient access to cheap capital—to be able to compete with the truly large companies.
The Labor Question
The other big problem that Uber will face with full autonomy is how to deal with its labor force. When the world switches from driver-assisted technology to full autonomy, a lot of people are going to be out of work and very angry.
I was in an Uber recently when a newscast started talking about Uber’s self-driving plans. The driver turned it up to listen to the discussion and expressed real dismay. The closer this gets, the more of a labor issue Uber is going to have on its hands. Its core advantage, which is labor access, will turn into a core disadvantage, a responsibility for helping its labor force find their next step.
In 2013, I wrote a post about how self-driving cars were likely to roll out. I suggested there would be three phases. The first will require a licensed driver—I suggested this would be 2018-2023. I expected the second phase would see a loosening of regulations over a 25- or 30-year period. And the final phase would be a fully autonomous transportation cloud.
As I expressed in that old post, there is a lot to look forward to. Transportation clouds are likely going to wreak havoc on the auto industry, drive direct car ownership toward zero, re-map real estate values globally and more.
What will be interesting is how the timing of the rollout in coming decades is going to materially impact who wins. Uber will continue pushing the technology envelope so it stays in the broader game. But if it is too successful with driverless cars too quickly, it will likely end up at a disadvantage against bigger competitors.
For now, Uber might be part of the coalition lobbying for the fully autonomous future. But it might be in Uber’s strategic interest to break from the pack and throw the brakes on true autonomy at some point in the not-too-distant future if they get the chance.