Volvo Cars and Nvidia: Up to 1,000 computing operations per second in the future!
At Volvo, they are once again ahead with new developments, all culminating in a sort of ultimate hyper-flexible modular system. The EX90 is the first software-defined model from the Gothenburg-based company. The large electric SUV is based on an industry-leading core-computing system: The onboard AI computer uses the Nvidia Drive SoC platform with up to 250 trillion computing operations per second. This high performance forms the foundation for all advanced activities that are meant to ensure a satisfying customer experience – from deep learning of AI-assisted active safety and assistance systems to safe autonomous driving in the future. CEO Jim Rowland, in an interview, left it open as to when and which level will be introduced here. Currently, according to the Scot, it applies until further notice: "Hands on the wheel, eyes on the road."
The company relies on both internal developments and intelligent partnerships with world-leading specialists like Nvidia for its software-defined vehicles to quickly offer the latest technology and make cars even safer.
To unlock the full potential of software-defined vehicles with core-computing architecture, Volvo Cars is now elevating its collaboration with Nvidia to the next level. According to Volvo, the integration of Drive Thor, which uses the Nvidia Blackwell GPU architecture, is expected to enable not only more advanced assistance and safety functions but also the further development of autonomous driving and the introduction of generative AI-based functions and experiences in the vehicle. Jim Rowan added:
"With Nvidia Drive Thor, our self-developed software will be better scalable across the entire model range. This improves the safety of future vehicles and guarantees top-class customer experiences while simultaneously reducing our costs and increasing our margins."
To further explore the potential of artificial intelligence (AI), Volvo Cars also utilizes Nvidia DGX systems through its own software company Zenseact – an AI supercomputing platform optimized for large workloads. It aims to foster the development of safe autonomous driving.
The DGX systems are initially used for training AI models before being deployed in future Volvo fleets on the road. With the platform, using specially developed AI infrastructure and optimized software stacks, the efficiency of training current and future AI models can be enhanced.
According to Volvo, the long-standing collaboration with Nvidia and the use of the Nvidia Drive platform provide "a deeper understanding of safety – parallel to the training of extensive AI models." These models are intended to help Volvo vehicles better understand the world around them. The goal is to further improve the safety and comfort of the vehicles and to advance autonomous driving.
"The Nvidia DGX AI supercomputer will enhance our AI training capabilities and turn our in-house AI training data center into one of the largest in the Nordic countries," says Anders Bell, Chief Engineering & Technology Officer at Volvo Cars, and adds:
"By leveraging Nvidia technology and building the data center, we can pave a quick path to a powerful AI that ultimately helps make our products safer and better."
What does that mean?
Volvo is taking the next big steps in software development and is working closely with Nvidia to do so. The vision of the Swedes remains a car that will no longer cost lives in accidents. A lofty goal, but according to Rowan, safety remains one of the four pillars of the brand's philosophy. And stronger chips can help with that. If they could first always correctly recognize and interpret traffic signs, that would be a significant step that everyone still needs to take….
Translated automatically from German.
Elektromobilität , Newsletter Elektromobilität , IAA Mobility , SUVs und Geländewagen , Hybrid , Antriebsarten, Kraftstoffe und Emissionen , Oberklasse- und Sportwagen , Carsharing , Autonomes Fahren (Straßenverkehr) , Ladeinfrastruktur , Verkehrspolitik , Formel E , Brennstoffzellen , Fahrzeug-Vernetzung und -Kommunikation , Fahrzeuge & Fuhrpark , Automotive-Messen & Veranstaltungen , Pkw, Kompakt- und Mittelklasse , Minis und Kleinwagen , E-Auto-Datenbank, E-Mobilität-/Automotive-Newsletter, E-Auto-Tests