Mercedes-Benz aims to close AI gap with China rivals through Wayve deal

Mercedes-Benz has struck a partnership deal with Wayve, the London-based autonomous driving start-up, as the German premium brand seeks to close an AI technology gap with Tesla and software-focused Chinese rivals.

The deal — Wayve’s third — is the latest effort by a traditional carmaker to keep pace with the industry’s costly transition towards more intelligent cars while seeking to retain control over key aspects of the software and vehicle design.

Wayve has already inked partnerships with Nissan and Stellantis, and the three automakers backed Wayve when it raised $1.2bn in February. The start-up also counts Nvidia, Qualcomm and Microsoft among its investors.

“I think there’s now a deep market conviction” about autonomous driving, Alex Kendall, Wayve’s co-founder and chief executive, told the FT. “I’ve seen automakers go from a year ago being almost sceptical . . . to now a deep belief that this technology works.”

Mercedes-Benz said it planned to integrate Wayve’s autonomous driving system into at least one model that it will deploy in the next two years, offering “advanced urban point-to-point driving assistance”.

Wayve’s current system delivers “Level 2” advanced driver assistance software — industry terminology for technology that requires active monitoring by a human driver behind the wheel, similar to Tesla’s Autopilot. Like Tesla, Wayve’s technology relies primarily on cameras and computers, which make its software more affordable.

Recommended

European carmakers are struggling to catch up with the heavy investments Tesla and Chinese rivals such as Xpeng are making to develop advanced autonomous capabilities.

But some in-house automotive engineers have been reluctant to hand over too much control of core technologies and data to tech giants, especially as software increasingly becomes the key competitive feature in a vehicle.

Wayve said its AI system allows carmakers flexibility in vehicle architecture and design, while Nvidia has also tried to allay concerns by allowing carmakers to customise its software stack, pitching its technology as open and transparent.

Mercedes-Benz is separately aiming to launch its new CLA model in the US, equipped with its own operating system and “Level 2++” capabilities using Nvidia software, by the end of this year.

Many carmakers have taken a multipronged approach, hedging their bets with different technology partners in various regions as well as in-house development of autonomous technology.

Nissan will combine Wayve’s software with its ProPilot semi-autonomous system and plans to deliver fully autonomous vehicles for Uber robotaxi services that use Wayve’s system.

But its engineers are separately developing its own “Level 4” system internally, which involves autonomous vehicles driving themselves with no human behind the wheel in limited, pre-mapped areas.

People close to the company said the Japanese group was divided into two camps: one enthusiastically embracing Wayve’s technology and another worried that the AI operated as a black box, creating liability issues.

A former senior Nissan executive said its in-house approach of rules-based self-driving systems would still be necessary for more advanced autonomous driving operations because the reason for an accident caused by a car using Wayve’s system could not easily be identified.

Kendall pushed back against black box concerns, saying there were tools in place to measure performance and validate capabilities, while its technology was also compliant with existing regulations.

“It will scale to a much greater degree than what a single automaker can do with an AI they try to build themselves. It’ll be . . . ultimately safer because of that model,” said Wavve’s chief executive.