Skip to main content
conference lunch move company map contacts lindholmen lindholmen 2 travel info


From Developer to Investor: Uber’s Self-Driving Metamorphosis

As of this week, Uber will no longer be in the driver’s seat when it comes to the pursuit of its autonomous driving ambitions. Instead, it will hand over its internal AD team to another competitor in return for a 26% stake. [1]

To “kill your darlings” is a common piece of advice given by experienced writers. You kill your darlings when you decide to get rid of an unnecessary storyline, character, or sentences in a piece of creative writing—elements you may have worked hard to create but that must be removed for the sake of your overall story. This week, Uber has just done the tech world equivalent of killing its darling by selling its self-driving development unit (Uber ATG) to another well-backed startup, Aurora Innovation. 

We’ve covered Aurora before at Drive Sweden Smart Mobility – Aurora Innovation was founded in 2016 by a former senior of Google’s self-driving program, Chris Urmson, who also made sure to recruit leaders within autonomy programs from Tesla and Uber. Its most notable investor is Amazon, which was part of a US $600 million funding round led by Sequoia Capital. The company also has non-exclusive partnerships with Hyundai and EV start-up Byton. 

This latest acquisition of Uber’s self-driving unit puts Aurora’s market valuation at US $10 billion. However, Uber ATG was valued at US $7.5 billion as recently as April 2019, following a large investment from Toyota. Since Uber only gets a 26% share of Aurora in return for the sale of its self-driving unit (plus they had to make an additional US $400 million investment in Aurora) this would mean that Uber ATG’s value had declined to about US $2.2 billion – a 70% fall in the space of just 18 months.   

The acquisition signals the end of a difficult few years for Uber’s foray into self-driving development. Uber has been operating in the autonomous space since 2015, initially in a partnership with Carnegie Mellon University and then forming their Advanced Technologies Group (hence the ATG in the name). This was followed by the purchase of an autonomous trucking startup called Otto in 2016. The death of a pedestrian in a March 2018 incident with an Uber ATG vehicle sent shockwaves through the whole self-driving car industry. Uber struggled to recover from the incident, with its vehicles taking almost a whole year to pass a series of closed track tests before returning to public roads again – just in time for the Corona pandemic. The New York Times attributes the plummeting revenues from Uber’s core ride-hailing business due to the pandemic as being the final straw behind the decision to sell Uber ATG. 


Personal comments

One of the ironies of this deal is that Uber got rid of its autonomous truck section in 2018 (perhaps related to legal troubles with Waymo) to focus more on the passenger vehicle side. However, it has now been purchased by Aurora, which differentiates itself from the rest of the self-driving field by including the development of automated trucks. It is hence unclear how this deal is strategic for Aurora. Specifically, how much of Uber’s self-driving experience with passenger vehicles which has mostly been based in urban environments (Pittsburgh, San Francisco, and Toronto) will be transferable or complementary to Aurora’s core focus of autonomous trucks operating on highways? Perhaps much of the value of the deal for Aurora stems from intellectual property developed during Uber ATG’s work in autonomous trucks (including its acquisition of Otto) before that was shut down. 

Seeing strategic value in this deal for Uber is a little bit easier. Self-driving cars were supposed to help Uber cut the biggest expense in its ride-hailing business – the labour costs of its drivers. But developing self-driving technology is itself expensive. This deal will mean Uber can stay in the self-driving game without needing to keep plowing more money into the venture quarter after quarter, year after year. By becoming an investor instead of a developer, Uber can still achieve its ultimate goal of operating a ride-hailing fleet free from any labour costs. It may just have to pay a licensing fee for the privilege – but by owning stakes in Aurora, Uber can possibly make the cost of this back and more. 

In the scheme of things, this is just another consolidation in the autonomous driving space. There will likely be many more before we finally get to a large-scale rollout of robotaxis and robotrucks.    


Written by Bobby Chen, RISE Mobility & Systems.



1. 2020-12-07. Uber, After Years of Trying, Is Handing Off Its Self-Driving Car Project