Images: Nuro
Nuro is betting that the autonomous vehicle industry will ultimately resemble the smartphone market, with a small number of technology companies supplying the underlying software and hardware that powers the majority of carmakers and mobility companies.
The California-based company initially set out to build and operate its own autonomous delivery vehicles. Now, it’s moved away from that strategy, instead seeking to license its self-driving system to companies like carmakers and ride-hailing operators.
Its partnership with Uber and carmaker Lucid is the most significant test yet of the strategy. Built on Nuro’s autonomous driving technology, the companies plan to deploy at least 35,000 self-driving vehicles across dozens of markets, with the San Francisco Bay Area expected to be the first commercial market and Houston following in mid-2027.
Nuro Chief Operating Officer Andrew Chapin describes the technology as “like an Android type operating system” for autonomous vehicles. In an interview with Zag Daily, Chapin says Nuro aims to be the “universal autonomy platform layer” for the industry.
The comparison is an attempt to separate autonomous driving from the rest of the transport business. Lucid supplies the vehicle, Nuro supplies the autonomous driving system and Uber provides the passenger network and operates the service around the fleet. Nuro’s technology is not yet authorised for paid, driverless passenger operations in California.
Breaking with the robotaxi template
Nuro was founded in 2016 by two early engineers on Google’s self-driving car project – which later became Waymo that same year.
The company first concentrated on purpose-built autonomous delivery vehicles but later decided to focus its business on autonomy rather than owning the vehicle and operating the service itself.
“If you are going to build a vertically integrated model where you have a consumer app, and perhaps you build the vehicle too and run the whole fleet operation – that is just massively, massively capital intensive,” Chapin says.
Nuro formally announced its shift towards licensing the Nuro Driver to automotive manufacturers and mobility providers in 2024 and says it has raised about $2.5 billion to date. Chapin has first-hand experience of the other end of the transport stack, as an early Uber employee who helped launch the London service before joining Nuro.
Nuro’s current model is to license its autonomy stack, including hardware and software, to partners that take on other parts of the business.
The model has precedent in automotive technology. Nvidia, the US chipmaker, has spent years supplying computing and autonomous driving technology to carmakers without building vehicles itself. Its Hyperion architecture is being adopted by companies including Lucid.
Nuro’s proposition is to supply the autonomous driving system itself, allowing different vehicle platforms to use a common autonomy system.
Rivals divide the work differently. Zoox, owned by Amazon, is building its own vehicle alongside its autonomous system and operating model. Waymo, Alphabet’s robotaxi business, runs both its driving technology and its taxi service, though Waymo doesn’t build its own vehicle.
Lucid’s separate deal with Bolt to deploy at least 25,000 autonomous vehicles across Europe follows a similar division of responsibilities. Lucid supplies the vehicles, while Bolt plans to own and operate the fleet itself. The vehicles are expected to use Lucid’s forthcoming Midsize platform and Nvidia’s Hyperion architecture, with Bolt setting the requirements for the autonomous vehicle.
A marketplace with room for few suppliers
Uber has increasingly sought to assemble a network of autonomous vehicle suppliers. Nuro is both a supplier to Uber and one of several companies competing for a place in that network.
“They want to be a marketplace of different AV providers,” Chapin says. “And I think that is well understood by us and by the marketplace.” He says Nuro is “totally happy to compete to win Uber’s business”.
He argues that the difficulty of reliable Level 4 autonomy, where a vehicle drives itself without human input in defined areas, will limit how many companies can offer it.
“The idea that a whole bunch of different companies are going to achieve that promised land of safe L4 is not one that I believe in,” he says.
For Nuro, the commercial attraction is that one technology could be licensed across many vehicle platforms and markets, cutting the investment needed for each deployment. Whether it can be adapted to new vehicles and geographies quickly enough to support large-scale commercial rollouts remains a central test of the model.
Nuro says its Driver has been integrated with 10 vehicle platforms and that its models have been trained on data from several countries, including Japan.
“When we first got a Lucid Gravity last year, it took us about seven weeks to get it driving autonomously on our close course,” Chapin says.
Seven weeks measures engineering adaptability. It does not show that the system can be rolled out quickly across a large commercial fleet which will need validation in different cities, road conditions and regulatory environments.
Taking out the safety driver
Nuro still has to demonstrate that its robotaxi system can operate safely without a human safety driver carrying passengers. The company received a California Public Utilities Commission permit in May allowing it to carry passengers with a safety driver and received a California Department of Motor Vehicles permit in April allowing driverless testing. Neither permit allows paid, driverless passenger operations.
“The gap between having a safety driver and taking the driver out is significant,” Chapin says.
Nuro plans to launch in the Bay Area later this year and in Houston in mid-2027. Its engineering fleet is approaching 100 Lucid Gravity vehicles across California and Texas.
The company says its validation programme combines road testing, closed-course testing and simulation and that it is approaching two million autonomous miles without an at-fault incident. That figure comes from Nuro’s own testing and includes its broader autonomous operations.
Nuro eventually expects the technology to reach privately owned vehicles though Chapin says commercial fleets and robotaxis will come first. For consumers, the economics differ because the cost of sensors and computing hardware has to be absorbed by a car that may spend much of its time parked and not gaining revenue.
“You’re not making this big capital investment as a consumer. You can take an Uber ride and experience a short trip in autonomy, gain confidence, gain trust in the technology,” he says.
The immediate opportunity, he believes, is to let consumers experience autonomous driving before they have to own it.
