The driverless race is on. GM’s acquisition of Cruise Automation was a shot from the starting pistol, and Uber’s move to buy Otto is just the first stretch of what’s going to be a marathon. Top talent is racing towards solving the hard technical problems ranging from computer vision to sensor fusion. Investor dollars are chasing driverless car startups. Public companies are seeking a driverless car angle to boost their share prices, and aren’t afraid to cut big checks. Meanwhile, consumers are anxiously waiting for their first ride in a robot car.

The big question is how? Will consumers walk into dealerships to purchase a robot car? Will driverless cars be manufactured by big automakers and operated exclusively by ride-sharing companies like Uber and Lyft? I expect driverless cars to be designed by the fleets that manage them.

Car companies will be enablers, but I don’t expect them to offer robo-cars to the public, though Uber is planning to be the first to pilot its robo-rides in Pittsburgh. The ride-sharing companies will have access to riders, but they aren’t built to be operators of bleeding-edge technology. The business of driverless cars will be in stark contrast to the GMs and Ubers of the world — making it the domain of startups for the following reasons:

When it comes to offering robo-rides, there’s a lot at stake

Big automakers have too much to lose. Ford and GM are each sitting at $50B market capitalizations. They are treated without mercy by public-market investors, who measure them by their profits and growth. It took Tesla over a decade to get to 15,000 electric vehicles per quarter. In comparison, Ford made $3.3B in profits selling cars that same quarter. Although building an electric car is no small feat, it is far less complicated than offering a fully-autonomous vehicle. As automaker CEOs are focused on delivering quarterly performance figures, what is their incentive to deliver fully-autonomous cars which will likely make up basis points of their overall fleet, and negative profits for decades to come?

Furthermore, automakers are in the business of selling consumer products. In fact, they have the misfortune of selling one of the most heavily regulated consumer products on Earth. The battery of emissions and safety requirements have made cars heavier and more complicated than ever. The distinctive qualities of iconic brands have been erased by the consolidation of suppliers and harsher regulations (perhaps the topic of another blog post). When defects occur, they cost the manufacturers billions. Toyota’s “unintended acceleration” almost brought the company to its knees in 2009. Audi was almost erased from the North American market in the 80s. The irony is that in both cases, there wasn’t actually a defect; it was a low threshold for human error. Toyota’s gas pedal was prone to being caught under a shifting floor mat, putting pressure on the accelerator as the mat was pushed forward over time by drivers who just let floor mats bunch up under their feet. In Audi’s case, there was little physical resistance required to release the vehicle from Park on the gear selector lever, making it easy for a driver to accidentally slide the transmission into gear and drive through their garage doors.

Driverless cars raise the stakes to an entirely new level. A driverless car runs on incredibly complex software where the slightest glitch can lead to catastrophic results. Although modern cars have ditched mechanical linkages for electronic couplings in braking, steering, and throttle control, the software that operates them is relatively simple. Furthermore, today’s vehicle software is not challenged with ethical questions around whether to risk its passenger or other motorists/pedestrians (see the classic trolley problem). In the inevitable situation where vehicle software fails, would an automotive company be willing to put its core metal-bending business at stake? Will automotive executives risk being reprimanded by their shareholders for drifting away from their core businesses by trying to operate as transportation service companies? More importantly, to what level should an automaker expose itself to the infinite liabilities that arise from building automated vehicles?

Car companies bend metal, they don’t offer rides

Automakers live and die by outproducing, outpricing, and outmarketing their cars — how the cars are put to use is none of their concern so long as they roll out of dealerships. Car companies have dabbled in service, especially finance and warranties, but most have pulled out to focus on their core business. Competitive advantage comes from fundamentals that date back to Henry Ford’s days: building and maintaining a supply chain, managing manufacturing operations, design, engineering, branding, and marketing. Developing cutting-edge robotics technology and managing a ride-hailing service is an entirely different endeavor altogether. Forward-looking car companies such as Ford, GM, Toyota, Volkswagen, BMW, and Mercedes are making sure to stay at the bleeding edge of development; however, they are not equipped, nor incentivized, to operate these vehicles. In fact, car companies have historically experimented with operating their vehicles, but quickly pulled back. Ford, for example, chose to purchase rental car company Hertz in the late-80s but opted out of the business in the early 2000s.

Ride-sharing companies do not manage assets or cutting-edge technology

The core appeal of ride-sharing companies is that they don’t carry assets on their books; they connect riders with drivers and take a cut. Enter driverless cars: ride-sharing companies can keep the 70% cut dished out to drivers; however, they are now responsible for the robo-cars that are hauling around their passengers. Are the ride-sharing companies equipped to manage this bleeding-edge technology? Are Uber and Lyft equipped to finance a fleet of complex machines, maintain them, run recharge stations, and address the inevitable malfunctions which could lead to catastrophic results? It is clear that for Uber, going driverless isn’t simply a matter of adding a feature — it’s a matter of changing the fundamentals of their businesses.

The solution: startups

At Lux, we are seeking visionary founders that solve big problems with unique technology. At the dawn of the automobile, many new types of companies were invented to bring humanity into the automobile age. Similarly, I believe that our driverless future will not be realized by either automakers, ride-sharing companies, or even a partnership between the two. I expect startups to identify the many novel challenges associated with driverless cars, such as low-cost lidar, sensor fusion, computer vision, modeling and algorithms, vehicle design, fleet management, tele-operations (if a robo-car gets stumped), charging stations, repair stations, in-vehicle entertainment and services, vehicle-to-vehicle communications, and of course, customer service. Perhaps one company will solve all of these, or maybe several billion-dollar companies will emerge from solving one of the most exciting challenges of our generation!