Xiaomi unveils 3-nm Xring D100 smart-driving chip for 2027 launch

Xiaomi unveils 3-nm Xring D100 smart-driving chip for 2027 launch

Xiaomi has announced the Xring D100, its first high-computing smart-driving chip built on a 3-nm process, capable of running 200-billion-parameter models locally, with commercial deployment targeted for 2027.

By CarsEVs Editorial Team

Source: cnevpost.com

Share

Xiaomi has unveiled the Xring D100, the company says is China's first high-computing-power smart-driving chip manufactured on a 3-nanometer process node.

The D100 is designed to run AI driving models with up to 200 billion parameters locally within the vehicle, eliminating or reducing the need for cloud-dependent processing for complex autonomous-driving tasks.

Xiaomi said the chip is targeted for commercial deployment in 2027, though no specific model or vehicle platform was named.

Domestic chip ambitions

The announcement marks a significant step in Xiaomi's efforts to develop in-house silicon for its intelligent-driving systems. The Xiaomi SU7 currently relies on Qualcomm and NVIDIA computing platforms for its driver-assistance features, making the D100 a potential cornerstone for future self-developed ADAS capability.

Several Chinese automakers and tech firms — including Huawei, Horizon Robotics, and NIO — have been pursuing domestically produced smart-driving chips to reduce reliance on foreign suppliers and keep pace with the rapid advancement of in-vehicle AI models.

Why 3 nm matters

Moving to a 3-nm process is a notable upgrade from the 5- to 7-nm nodes common in many automotive-grade SoCs today. A smaller node generally delivers higher performance per watt, which is critical for running large language and perception models inside a vehicle without excessive power draw or thermal output.

The 200-billion-parameter figure places the D100 among the most compute-hungry automotive chips being discussed in China's EV sector, where end-to-end autonomous-driving models are becoming the industry norm.

Full specifications, including inference performance, memory configuration, and power consumption, were not disclosed.

Privacy settings

Choose which cookies CarsEVs may use. Your choice is stored on this device.