Wayve’s deal with CoreWeave shows that compute capacity is becoming as decisive as data in the race to commercialise driverless cars
Wayve will train the foundation models behind its AI Driver on CoreWeave Cloud under a multi-year agreement, running on Nvidia Blackwell graphics processors. The capacity covers training, evaluation and simulation, and is meant to speed the autonomous driving developer’s route to commercial deployment.
The AI Driver learns from real-world data gathered in more than 70 countries and is aimed at uses ranging from privately owned cars to robotaxis. Wayve says that workload needs reliability over long training runs and enough visibility to show each model version beats its predecessor and to validate safety at scale.
Wayve already used Weights & Biases before expanding onto CoreWeave Cloud. Because its mapless, hardware-agnostic system is offered to automakers and fleet operators, how quickly it can retrain and verify models bears directly on how fast new versions reach vehicles.
“Our AI Driver gets better by running this data flywheel: gathering real-world driving experience, training on it, evaluating it, and deploying only what outperforms what is already on the road,” Silvius Rus, Senior Vice President of Engineering at Wayve, said. “Being able to access CoreWeave’s AI cloud, purpose-built for workloads like ours, helps accelerate the commercial deployment of our AI Driver.”
Source: Wayve
Source: www.automotiveworld.com
