Last March, just as the Washington, DC, area was locking down for COVID-19, I traveled to George Mason University in Fairfax, Virginia, to eat a hamburger delivered by a robot.
The robot was owned by delivery startup Starship. Today, Starship announced that it has completed 1 million deliveries. Annika Keeton, a college student in Bowling Green, Ohio, was Starship’s millionth customer.
In recent years, a lot of hype has followed companies like Waymo and Tesla as they try to build full-size, go-anywhere self-driving cars. But designing these vehicles has proven fiendishly difficult. Tesla’s self-driving software still requires careful supervision, while Waymo’s driverless taxis are still limited to one corner of the Phoenix metropolitan area.
Starship chose to tackle a much easier technical challenge. Its vehicles never go faster than four miles per hour (6.4km/h), dramatically reducing safety concerns. Starship has focused on offering service on college campuses, a relatively controlled and forgiving environment for autonomous robots. As a result, Starship was able to launch commercially a few years ago and expand rapidly.
The company reached 100,000 deliveries in August 2019 and 500,000 deliveries in June 2020. It expects to complete another million deliveries in a matter of months.
Obviously, Starship isn’t an immediate threat to Waymo. But self-driving startups exist on a spectrum between Starship and Waymo. There are several startups—including Nuro, Udelv, and Gatik—working on driverless delivery services on roads rather than sidewalks. These vehicles will go faster than 4mph, but because cargo isn’t in a hurry, the vehicles can go relatively slowly. Nuro’s custom delivery robots, for example, have a top speed of 25mph (40km/h).
There are also a number of startups working on driverless taxi and shuttle services that operate at low speeds in limited areas. Voyage is building a driverless taxi service at a large retirement community in Florida where the top speed is 25mph. Optimus Ride partners with large commercial landlords to provide shuttle services in planned communities. May Mobility is running a self-driving shuttle service in Grand Rapids, Michigan.
Obviously, if Waymo can get its full-speed self-driving technology working at scale, it can easily enter these narrower markets. But if it takes time for Waymo (and other high-end robotaxi companies like Cruise, Argo, and Zoox) to scale up—for logistical, technical, or regulatory reasons—that will leave an opening for more targeted services to grab market share. And once these startups are operating at a large scale, they may be able to improve their technology to handle higher speeds and more demanding situations, allowing them to more directly challenge companies higher on the self-driving food chain.