The Chinese electric vehicle (EV) manufacturer NIO is stepping up its endeavors in developing in-house chips, aiming to reduce its reliance on NVIDIA and boost profitability. On April 24, NIO's CEO, William Li, announced that the company's self-developed chips are tailored to seamlessly integrate with its proprietary algorithms and sensor setups, significantly enhancing efficiency in artificial intelligence applications, notably Advanced Driver-Assistance Systems (ADAS).

NIO has successfully introduced the 5nm intelligent driving chip, dubbed 'Thor NX9031', and the lidar main control chip, named 'Yangjian NX6031'. Notably, the 'Thor NX9031' stands as the world's inaugural mass-produced automotive-grade 5nm high-level intelligent driving chip, boasting over 50 billion transistors. A single unit of this chip delivers computational performance on par with four NVIDIA Orin-X chips. To date, NIO's cumulative shipments of self-developed chips have surpassed 550,000 units. These chips have been deployed across several models, including the ET9, ES9, and ES8, and are slated to be incorporated into the 2026 L90 model under the Onvo brand, which will cater to the RMB 200,000-300,000 price range.

In a strategic move, NIO has spun off its chip division, 'Shenji', into an independent entity and is now open to supplying chips to external clients. Li underscored the significance of self-developed chips for NIO's enduring global competitiveness. He highlighted that these chips are instrumental in establishing a closed-loop system encompassing 'data-algorithm-computing power', fostering profound synergy between hardware and software, and ultimately elevating the company's overall profitability.