Horizon Robotics targets mass rollout of city-level driver assistance to China’s 100,000-yuan car segment

Shanghai (Gasgoo)- Horizon Robotics is preparing to bring advanced urban assisted-driving capabilities to far more affordable vehicles in China.

Speaking at the company's 2025 Technology Ecosystem Conference on December 8, CEO Yu Kai said models equipped with Horizon Robotics' city-level ADAS, powered by a single Journey 6M chip, are close to entering mass production. The move is set to push features once reserved for premium cars into the mainstream 100,000-yuan price bracket.

According to information shared at the event, Horizon Robotics will initially commercialize its Journey 6M–based system through two cooperation models. One pathway focuses on toolchain collaboration with partners such as Bosch, ZYT, and QCraft, enabling automakers to integrate the chip into their own software stacks. The second approach centers on providing packaged algorithm services, working with companies including DENSO, CARIZON, and neueHCT to deliver ready-to-deploy solutions for vehicle programs.

Horizon Robotics also unveiled its fourth-generation BPU architecture on December 8, named "Riemann" in tribute to the mathematician. The new architecture introduces improvements in high-dimensional data processing, algorithm optimization, and overall system efficiency. The company says Riemann delivers stronger compute performance, more diverse operators, and significantly better energy efficiency. It will debut on the upcoming Journey 7 series of automotive AI chips.

Yu described the BPU architecture as an "orchestra" of computing elements that must work in perfect coordination to support the next wave of automotive AI. He outlined the evolution of Horizon Robotics's chip platforms—from Bernoulli to Nash and now to Riemann—highlighting how each generation has been designed to match the rising computational demands of autonomous-driving algorithms. The latest architecture, he said, is now capable of supporting full-scenario navigation-on-autopilot and real-time interactive decision-making.