Uber has expanded its operating zones for autonomous vehicles (AV) in Austin with Waymo and Abu Dhabi with WeRide, and has also launched with Waymo in Atlanta. But "I think the commercialisation is going to take time," Dara Khosrowshahi, CEO of Uber, told analysts. The "Austin launch continues to go really well in terms of utilisation. Atlanta launch has been -- it's early, but the Atlanta launch has been great. And in both cases, the average Waymo is busier than 99% of our drivers in terms of completed trips per day... the focus now is how do we bring this product to market as quickly as possible because it looks like from a consumer standpoint and from a safety standpoint, it's a real hit."
Uber said in July that is is aiming to deploy at least 20,000 AVs made by Lucid equipped with the Nuro Driver over six years. The vehicles will be owned and operated by Uber or its third-party fleet partners and made available to riders exclusively via the Uber platform. The first Lucid-Nuro robotaxi prototype is operating autonomously on a closed circuit at Nuro’s Las Vegas proving grounds.
"I do think that you will see us do more deals like Nuro, Lucid in the early days of autonomous as we're proving out the economics of the marketplace," Khosrowshahi told analysts. "Once we prove out the revenue model, how much these cars can generate on a per day basis, there will be plenty of financing to go around, third-party financing. We've talked to private equity players. We've talked to banks, et cetera. And while it will take some time, we're very confident that these assets are going to be financeable."
Affordable hardware is the bottleneck
The Uber CEO went on to say that the development of AV software has been accelerated by the arrival of larger AI models. "There are some approaches that are kind of perception models interpreting the world and then prediction models that decide what to do. Some players like Tesla and Wayve, for example, are going with single models. But whatever the approach is, it is really accelerating in terms of time to market. And we actually think the harder part of commercialisation is going to be hardware, bringing on hardware platforms and hardware partners that can build these platforms at scale affordably because today, at least AV vehicles that are able to be deployed at scale are pretty expensive."
However, he noted that Baidu has established hardware platforms that "look to be effective at scale and quite affordably.... The Apollo Go for is one example of that....And then in terms of Tesla, listen, we see them on the streets right now, the deployment that we're observing is very, very small."
Uber has identified three potential business models - the merchant model (in which Uber will pay the EV partner a set fee per trip or per day), the agency model, which is a revenue share, and a licensing model in which Uber buys the vehicles and then licenses the software either by the month or by mile. "I think you'll see all three of those models in our marketplace over the next five years," Khosrowshahi explained.
Uber's drivers are involved in data collection to support the development of autonomous vehicles. "It's not really something that I'm looking to make profits on but really helping AV get to market faster," Khosrowshahi noted. "And then one area that we're really excited about is kind of one way to look at Uber is, it is a platform for work. We've got almost nine million people who are earning on our platform, and they can earn in different ways other than driving or transporting things. An example of that is actually our Uber AI solutions...they are engaging in data labelling, translation, map labelling, tuning algorithms, et cetera. And these are sometimes drivers on our platform, sometimes kind of specialists that we bring on to the platform, but it's using the core Uber capability, which is sending out tasks to earners all over the world." Source: Uber earnings transcript
