Oct 28, 2021: The Information’s Creator Economy Summit

Autonomous Vehicles Uber/Lyft

Uber Talks to SoftBank-Backed Nuro About Autonomous Food Delivery

Uber has had discussions with SoftBank-backed self-driving vehicle developer Nuro about using Nuro vehicles to help automate Uber’s hot food delivery service Uber Eats, according to a confidential document reviewed by The Information. Such a deal could someday help Uber reduce losses on Uber Eats, a critical issue for investors as the ride-hailing firm goes public.

The discussions, which center on Uber Eats testing Nuro vehicles in suburban Houston later this year, show how Uber is pursuing long-standing hopes of using self-driving technology to cut costs of hiring drivers. The technology could be hugely valuable for the ride-hailing firm’s long-term profitability, although prospects for self-driving cars remain highly uncertain.

The Takeaway
• Nuro vehicles would take food from restaurants to central hubs
• Nuro’s vehicle software makes frequent mistakes
• Proposed effort does not involve Uber’s autonomous vehicle program

In the case of Nuro and Uber, the idea under discussion is that Nuro vehicles would carry Uber Eats deliveries for the beginning of their trip to customers, what is known as the “first mile” of a delivery. That could improve Uber Eats’ margins by as much as 20 percentage points in some markets, if it has enough scale, according to the document.

Earlier this year SoftBank agreed to invest $940 million in Nuro, which was founded in 2016 by former key members of the Google self-driving car team. But Nuro’s discussions with Uber—which is also partly owned by SoftBank—began before the SoftBank investment. SoftBank is putting the money into Nuro in stages, said a person close to the company. A deal with Uber could help Nuro prove how it can be a business even if it doesn’t handle a delivery from end to end. Nuro would get a cut of transactions it helped deliver. (Nuro already is testing delivering groceries for the Kroger supermarket chain in Arizona.)

The Uber-Nuro discussions follow a previously announced partnership agreement between General Motors’ Cruise self-driving car division and DoorDash, which is Uber Eats’ top rival. (SoftBank is a major shareholder in both those companies as well.) But in that case, the announced deal is an apparent attempt to automate an entire food delivery trip, with the vehicle carrying food from just outside the restaurant to the location of the customer. Another food delivery app, Postmates, which also is prepping for an IPO, has said it is building its own small delivery vehicle that would roam on sidewalks as opposed to streets.

The Uber-Nuro partnership would work like this: Restaurant workers would take an order and place it in Nuro’s experimental vehicles on the street. Those vehicles would drive themselves to a central location, from which Uber Eats’ human drivers would pick up the orders and drive them to people’s homes as they do today. The hope is that the centralized hub for orders would allow drivers to handle more food orders than they currently do and potentially make more money because they won’t have to spend time going to each restaurant to pick up the food.

Uber Eats was used by at least 15 million customers in the fourth quarter of last year and generated $3 billion in gross revenue in the first quarter of the year, Uber said. But Uber Eats shares most of that money with restaurants who supply the food and the drivers who deliver the meals. After those payouts, Uber said it would book at least $218 million in net revenue from Uber Eats, up from $183 million in the same period a year ago. But Uber said it tripled the amount of financial incentives for drivers to get that growth. Partly as a result of those costs, Uber Eats loses money.

Nascent Technology

To be sure, the technology underlying Nuro’s vehicles likely is years away from being reliable in complex driving scenarios, and its ability to produce vehicles at scale will take time too. (Nuro’s robot prototypes are narrower and shorter than a typical car and are monitored remotely by Nuro employees.) The company has been testing such vehicles, as well as using its autonomous driving software to power normal Toyota vehicles, in the suburban Phoenix area. The software frequently makes mistakes, said a person close to Nuro, as human backup drivers regularly take over the wheel of the Toyotas in order to prevent a potentially unsafe situation.

These challenges, which are common in the field, are part of the reason why Nuro and Uber are focused on automating only the first mile of deliveries rather than the full, several-mile-long delivery. The Uber document about the project is labeled as coming from the “Office of the CTO,” with Uber CTO Thuan Pham named alongside two Uber engineers, Waleed Kadous and Steven Choi. A spokesman for Uber did not make them available for comment and said the company would not comment.

Dave Ferguson, CEO of Nuro, said in an email: “While we have a great deal of respect for other companies in the delivery ecosystem, including Uber Eats, we have no existing partnership with them, nor any other information on partnerships right now.”

Based on the document regarding the proposed Uber-Nuro deal, the partnership does not appear to involve Uber’s own autonomous vehicle development unit, which is focused on trying to make robotaxis to aid Uber’s core business.

Robotaxi efforts, including Uber’s, have faced persistent technical problems and challenges in transporting human customers. (See: Waymo’s experience in Phoenix.) That’s why delivery companies and automated trucking firms, whose cargo isn’t human, have gotten more attention lately among investors and talent in the field.


Amir Efrati is executive editor at The Information, which he helped to launch in 2013. Previously he spent nine years as a reporter at the Wall Street Journal, reporting on white-collar crime and later about technology. He can be reached at [email protected] and is on Twitter @amir