In response to soaring procurement costs and an intensifying AI computing power race, Chinese electric vehicle manufacturer NIO is accelerating its efforts to reduce reliance on NVIDIA. The company recently announced that its new sub-brand, Onvo, will fully adopt its self-developed "Shenji NX9031" intelligent driving chip, aiming to break free from dependence on overseas giants like NVIDIA and gain control over its supply chain.
Self-developed chips have become a key tool for cost reduction. High-end intelligent driving chips have long been dominated by giants like NVIDIA, leaving automakers with limited bargaining power. NIO founder William Li revealed that during peak procurement periods, the company paid up to $300 million annually to NVIDIA, with chip costs per vehicle reaching approximately 11,000 yuan.
"From the perspective of using R&D to lower costs and improve gross margins, developing our own chips is certainly cost-effective," Li stated. With the mass production and deployment of the self-developed "Shenji NX9031" chip, NIO has successfully reduced chip costs per vehicle by about 10,000 yuan. Based on current annual sales of several hundred thousand units, this move is expected to save the company over 1.8 billion yuan in expenses each year.
The "Shenji NX9031" is the world's first mass-produced 5-nanometer automotive-grade intelligent driving chip, boasting aggressive specifications. Reports indicate that a single Shenji chip delivers computing performance equivalent to four NVIDIA Orin-X chips and contains over 50 billion transistors. The chip's memory bandwidth and image processing capabilities have been optimized for large model requirements, designed to meet the complex computational demands of next-generation autonomous driving algorithms.
In the 2026 Onvo L90 model, the Pro version will feature the NVIDIA Orin X solution, while the higher-tier version will upgrade to the self-developed "Shenji NX9031" chip along with a lidar system. This tiered strategy demonstrates NIO's control over its supply chain during the transition period.
Developing chips in-house is not solely about cost reduction but also about securing "iterative certainty" in the AI era. Li pointed out that intelligent driving chips evolve extremely rapidly, and "without offering the latest technology, you cannot compete in the market."
Mastering the underlying chip technology helps NIO create a closed loop integrating "data-algorithm-computing power," enabling deep synergy between software and hardware. Despite cumulative R&D investments exceeding 67 billion yuan and facing the strategic goal of achieving profitability by 2026, NIO has chosen to spin off its chip subsidiary Shenji and seek external financing to adopt a more flexible approach in this critical "AI turnaround battle."
Comments