Volvo makes a bigger bet on Nvidia

Volvo is significantly increasing its investment in autonomous driving technology through a strengthened collaboration with Nvidia. The companies revealed on Thursday that Volvo will incorporate three components of Nvidia's hardware and software into its future vehicles, enabling the cars to navigate both urban streets and highways autonomously.

Volvo has not disclosed the monetary value of the partnership or the specific vehicle model that will be the first to utilize this advanced Nvidia technology stack. However, Alwin Bakkenes, Volvo's head of software development, offered insights into the scale of the deal and Volvo's broader strategy. He emphasized that traditional automakers typically operate in a fragmented manner, relying on various suppliers for different components.

In contrast, Volvo has undergone a transformation, now functioning as a complete technology stack company. Bakkenes further revealed that the partnership will commence with a new vehicle model set to launch in the near future, with other vehicles following within the next decade.

Interestingly, while Volvo's association with Nvidia is not entirely novel, this expanded partnership signifies the company's commitment to integrating cutting-edge autonomous driving software into all its vehicles, regardless of whether they are electric or gasoline-powered. Volvo's battery-electric vehicles currently employ Nvidia's Orin computer chip, which utilizes data from cameras, radar, and other sensors to enable the vehicle to comprehend its environment and make appropriate driving decisions.

According to Bakkenes, Volvo plans to standardize even more on Hyperion, another Nvidia offering, particularly in terms of sensor sets to broaden the technology stack's reach.

The expanded partnership entails Volvo merging its vehicle safety systems and engineering expertise with Nvidia's Hyperion sensor and compute architecture, Alpamayo driving software stack, and Halos safety evaluation layer. Alpamayo, a suite of AI models, datasets, and simulation and development tools for autonomous driving software, is designed to be redundant and diverse, according to Ali Kani, Nvidia's automotive division head.

Kani explained that the stack combines Alpamayo's AI models with classical software that leverages machine learning and camera and radar perception. This approach, he emphasized, enhances safety by employing multimodal sensors. Furthermore, Kani noted that the open-source Alpamayo software has garnered over 650,000 downloads from developers worldwide.