Xpeng to Deploy Proprietary Autonomous Driving Chip Rivaling Nvidia Thor
Chinese electric vehicle manufacturer Xpeng Motors plans to accelerate the deployment of its proprietary autonomous driving chips, aiming to integrate them into an upcoming car model within the current quarter. This timeline marks a significant shift from the previously scheduled second-half rollout, according to CEO He Xiaopeng. The new semiconductor, named 'Turing,' is reported to deliver approximately 700 trillion operations per second (TOPS), placing its performance on par with Nvidia’s next-generation DRIVE Thor chip. Furthermore, the chip features neural processing units capable of training large language models with over 30 billion parameters, enhancing the vehicle's AI reasoning capabilities. He Xiaopeng indicated that these chips will eventually be utilized in Xpeng vehicles globally, pending the opening of relevant international policies. The announcement was made ahead of Xpeng’s Global Brand Night in Hong Kong, highlighting the company's strategic move toward vertical integration in critical automotive technologies. This development underscores the intensifying competition in the autonomous driving sector, as major EV makers seek to reduce reliance on external suppliers like Nvidia by developing in-house silicon solutions that offer competitive performance and potentially lower costs.
Wire timeline
Xpeng to Deploy Proprietary Autonomous Driving Chip Rivaling Nvidia Thor
Chinese electric vehicle manufacturer Xpeng Motors plans to accelerate the deployment of its proprietary autonomous driving chips, aiming to integrate them into an upcoming car model within the current quarter. This timeline marks a significant shift from the previously scheduled second-half rollout, according to CEO He Xiaopeng. The new semiconductor, named 'Turing,' is reported to deliver approximately 700 trillion operations per second (TOPS), placing its performance on par with Nvidia’s next-generation DRIVE Thor chip. Furthermore, the chip features neural processing units capable of training large language models with over 30 billion parameters, enhancing the vehicle's AI reasoning capabilities. He Xiaopeng indicated that these chips will eventually be utilized in Xpeng vehicles globally, pending the opening of relevant international policies. The announcement was made ahead of Xpeng’s Global Brand Night in Hong Kong, highlighting the company's strategic move toward vertical integration in critical automotive technologies. This development underscores the intensifying competition in the autonomous driving sector, as major EV makers seek to reduce reliance on external suppliers like Nvidia by developing in-house silicon solutions that offer competitive performance and potentially lower costs.
TechNode