A World of Driverless Cars (and Opportunity)
We believe this sector will have everything needed to make lucrative investments. Ronny Conway

With a number of recent high-profile controversies, discussions about the future of Uber are rampant across the technology world. Considering this, I thought it would be a good time to weigh in on one area where we believe Uber is heading in the right direction: the future of autonomous transportation. From an investment standpoint, we think of autonomy in terms of land, sea, and air, but for the purposes of this essay, I want to focus specifically on self-driving cars. We believe this sector will have everything needed to make lucrative investments.

The market is massive, with all of the major tech companies (Apple, Google, Intel), ride-sharing companies (Uber, Lyft, Didi), and major OEM’s (Ford, Tesla, GM) around the world going into this space. We estimate the combined market cap of these companies is at least four trillion dollars, and they all will need to have some strategic exposure to the space. And there are a number of layers within the developing technology stack that companies can attack.* On the software side, with the advent of AI and computer vision, there is a path to build transformative products in the space. While Google has done the best job of building the autonomous car so far, the founding teams we’ve met from external companies in this space are off the charts. From a pure background perspective, we would like to back all of them. These are some of the smartest people we’ve ever come across in our investing careers and they are all building valuable IP. So from a market, product, and team perspective, the space is producing a lot of investment grade opportunities.

So where will the best opportunities arise? First, we have to look at the car itself. We believe the self-driving car market will resemble the new car market today, where very little money is made on the automobile. This is because multiple companies will be producing autonomous cars (at the very least Ford, Tesla, GM), causing prices to quickly race downwards. This will lead to the commoditization of the self-driving car. But many deep-pocketed companies will need to own the valuable software IP powering those cars and that’s where we think there are some interesting opportunities.

Currently, the biggest “new” market in automotive software is ride-sharing. Uber and Lyft’s big advantage here is their network effect, putting drivers and riders together as fast and cost-effectively as possible. It’s a classic marketplace where there is defensibility and some price flexibility/control based on liquidity. There is a true network effect here today, but when the driver is removed from the equation and replaced by an autonomous car, the business begins to look a lot different. Even more so considering fewer people will likely own cars themselves when they can hail a self-driving car on-demand. So in this not-too-distant-future, how can a company begin to create network effects? We believe it will be through routing and capacity. As with the new car market, consumers will be price conscious and will most likely choose the cheapest ride-sharing option. We believe that capacity and routing will be the key factors that drive prices down.

I want to explain a bit more about what I mean by capacity and routing. In a world where self-driving cars become a commodity, any car will have the ability to deliver a person from point A to point B for a fixed price. That price will most likely be determined by an equation such as

the cost of the car + fuel + maintenance
# of riders

This is why LyftLine and UberPool are such important initiatives to these businesses. Due to the reliance on raw materials, it will be very difficult to gain a strategic advantage by driving down the cost of the car, fuel, or maintenance. We think the “# of riders” is the variable where the strategic advantage will kick in. If I have the most riders on my network in a certain geo, and I have the software to handle that demand, then I can route the most efficiently. If a given ride costs $10, that ride can be split between two people for $5 each or between three for $3.33 each. In this case, riders will always pick the $3.33 fare. Today, with UberPool and LyftLine, the leading ride-sharing companies are already gathering the data and customer loyalty needed to build the most efficient routing and capacity methods. None of the other big tech companies or OEM’s have this data, providing Uber and Lyft with a massive advantage. For instance, unlike Uber, Ford today does not know that during rush hour, going one block out of the way to pick up another rider (assuming a nominal decrease route in efficiency), each rider will pay X dollars less. This example dramatically conflates how hard of a problem this is, but also hopefully illustrates the data advantage only a few companies have. Without the actual data and the ability to predict, even if you have the self-driving car itself, it will be very hard to create another ride-sharing service from scratch. Prices will drop too quickly on the ride-hailing side and the cost of a self-driving car will be too great to begin building another network and gathering the data needed to have a price competitive offering.

Google’s data play here is Waze, which is a great indicator of where traffic is going, but it cannot accurately predict how many people will want to be picked up and dropped off at specific locations at a given time on a given day. Tesla faces a different problem. Elon Musk wrote a fantastic post called “Master Plan, Part Deux” where he outlines how Tesla owners will some day be able to make money off of ride-sharing by dropping you off at your destination, driving passengers around until you need to be picked up and driven to your next destination, effectively using the excess capacity of your car. But I don’t know how that can be achieved unless Tesla can accurately predict capacity AND routing AND compete on price. The OEM’s will surely target this market, and I don’t want to rule them out, but in order for them to make this ride-sharing jump, they’ll have to go from being product companies to service companies, which I just don’t believe they can do. Where they come into play is the car manufacturing itself, and they are very good at it. The technology world drastically underestimates how good they are at producing quality cars at a cheap price and how difficult that is. We believe there will be partnership opportunities for them to sell into ride-sharing networks, and that is where they will be able to make money in this new world.

So where are the investment opportunities? As I laid out above, we don’t believe there will be any full stack new self-driving car companies because (a) there are too many companies doing various parts of the stack really well, (b) everyone is chasing Uber and Lyft for ride-share data, and (c) it’s going to be really hard to create a successful new car manufacturing company from scratch. Where we think we can play is the software stack in the car. Whether it be a team out of Stanford that figures out how to build lidar really cheaply or brilliant computer scientists who have a machine vision breakthrough, both are examples of specific parts of the autonomous car stack that will be valuable long term but will also be something core to our roots: software. Since we won’t be targeting full stack investments in this space, the startup costs for the companies we look at resemble that of a traditional software startup, which obviously plays well in our model.

All of the deep-pocketed technology companies and ridesharing companies are going after the self-driving car space, but if you believe price in the autonomous ride-sharing world is what will prevail (driven by capacity and routing), then you should believe Uber still has a huge leg up on everyone. Given Uber’s 80% market share, it will be very hard for anyone to beat them, but as we all know, you should never crown a king too early or rule out any new startup from drastically changing the market.


*To name a few: car production, services, safety and security, autonomy, infrastructure, advanced manufacturing, on-board sensors, and specialty vehicles. For more, see the Wired article “Mapped: The Top 263 Companies Racing Towards Autonomous Cars”.