How successful AV development might play out

This year, rumors that Google parent Alphabet was seeking to sell a portion of its investment in Waymo confirmed that the journey to Level 5 autonomy will not be direct or quick. And it will be grindingly expensive — even for Alphabet's seemingly bottomless purse. More than a decade of "it's coming soon" has given way to underappreciated, steady technical progress for autonomous vehicle deployment.

According to Amara's law (coined by Roy Amara, Stanford University computer scientist and past president of the Institute for the Future), "We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run." The prevailing assumption that the current AV leaders will continue to dominate the market underestimates the profundity of the changes that autonomous vehicles will bring over the next decade.

Particularly because AVs are linked with electric vehicle development, the industry faces enormous long-term capital expenditures and uncertain technical success. AV development resembles high-stakes aircraft development, where companies wager their existence on complex, unknowable technology.

We sketch below our vision for successful AV development, including an accounting of the challenges that would-be industry leaders must conquer. Success will require capital and fortitude, but ingenuity will be the real differentiator.

Upfront costs are high, and delays and setbacks are routine, yet sources for revenue (let alone profit) are obscure. Because AVs will be dramatically more expensive and simultaneously more readily shared and less prone to collisions, new-vehicle sales will continue to plummet. Even if private car ownership persists, households may choose ride-sharing over owning a second or third vehicle.

Developing interim revenue streams from AV technology is imperative. Current AV technology has practical applications that should be commercialized to defray AV development expenses. For example, Waymo is planning to sell one type of its proprietary lidar units to non-AV companies. Similarly, Mobileye's advanced driver-assistance systems are compatible with non-AV vehicles and provide revenue as its AV efforts mature.

Toyota, with its Guardian and Chauffeur systems, has taken a two-pronged approach: It can make money by deploying ADAS and gain valuable experience to develop the Chauffeur system. Look for ride-sharing companies to investigate selling data to and about passengers (with consent) to subsidize rides.

Will hardware and software production be integrated? Members of the industry are partnering with one another. Waymo relies on Chrysler, Porsche and Jaguar for its hardware. Mobileye cooperates with all major automakers and Chinese search engine giant Baidu. Honda and General Motors have tie-ups. Aurora is partnering with Volkswagen and Hyundai. Winners will assemble joint ventures where each participant contributes different strengths, including players such as Amazon or Kroger from outside the industry.

Faced with a global shortage of skilled scientists, AV companies must compete fiercely for talent, both with one another and other industries.

Aided by California's hawkish prohibition on noncompete agreements, employees have moved freely within the industry, taking invaluable expertise with them. Famously, Waymo's rich payments to its AV developers have funded many of its competitors, including innovators such as Nuro and Aurora. Innovative approaches, such as the use of equity to bind and incentivize valued employees, as Argo.ai has pioneered, are required.

Technical progress will continue as AI systems, sensors and mapping improve. At the same time, simpler, more focused application of AVs will fuse immediate benefits with long-term payoff. Voyage's retirement community shuttles, May Mobility's urban trolleys and Optimus Ride in the Brooklyn Navy Yard in New York are less ambitious than the fleets of AV taxis that automakers and Waymo are planning, but they are also ready to deploy, which speeds learning and moves a step closer to revenue.

To the same end, commercial AV uses are promising. TuSimple has implemented long-haul AV freeway operations as it tackles local AVs. Low-speed, geography-limited grocery delivery systems such as Nuro are also likely to make an immediate impact. Competitors must also find ways to simultaneously develop their AV technology and earn revenue. By starting small, they also earn confidence from regulators and the public.

Combining AVs with simplified EV drivetrains means that maintenance costs will plummet. Vehicles' forms — their size, look and function — may become substantially more diverse. AVs may be designed for more specific roles than traditional cars, such as Voyage's shuttles, but cultivating the flexibility in vehicle production will be the key.