Volkswagen to develop in-house smart driving chip with Horizon Robotics
Volkswagen has made significant commitments to Chinese semiconductor and AI technologies to bolster its smart car capabilities, especially in the Chinese market, where competition is fierce from local rivals who are being supported by the Chinese government.
The company has now made two new announcements heralding developments that are no doubt geared towards giving Volkswagen a better fighting chance in the Chinese market.
On Wednesday, Volkswagen revealed that CARIZON, its joint venture with China’s Horizon Robotics, is working on developing its first in-house chip to power the smart driving capabilities to be equipped in the next generation of cars it will release in the Chinese market.
The company says the chip will be useful for processing data from cameras and sensors in a car with a single-chip computing power of around 500 to 700 tera operations per second (TOPS). The product can be expected to hit the shelves within the next three to five years.
It is putting the chip on the same level as some of Nvidia’s, like the latest Thor chip, which is popular in premium Chinese EVs.
Volkswagen to develop a chip
The new chip is reportedly designed specifically for the complex road conditions in China and will draw on capabilities like real-time decision-making, high computational efficiency, and adaptation to navigate diverse urban scenarios.
Volkswagen claims the move will deepen research and development capabilities. However, it has failed to elaborate on whether the chip will be used in cars to be sold outside the world’s top auto market.
The move aligns with Volkswagen’s recent actions in China, its largest market, where it has been trying to make progress by improving product competitiveness with technologies developed by or in conjunction with native Chinese companies. The moves are part of steps to close the gap between it and the fast-moving local rivals.
If it works according to plan, the chip could help Volkswagen regain a larger share of its market from rivals while encouraging further integration into the Chinese ecosystem.
Volkswagen has also partnered with XPeng
According to an announcement made today at XPeng’s AI Day 2025 event, Volkswagen has been officially confirmed as the first strategic partner for XPeng’s second-generation VLA (Vision-Language-Action) system and Turing AI chips.
The development reinforces Xpeng’s expanding ecosystem, which now covers autonomous driving, personal air mobility, and humanoid robotics similar to Musk’s Tesla. This marks the sixth collaboration between the two companies and comes after earlier joint work on electrical architecture systems and ultra-fast charging networks.
Volkswagen is expected to use XPeng’s self-developed Turing AI chip in the manufacture of its upcoming Chinese-market vehicles and with the announcement, XPeng, primarily an EV maker, has emerged as a chip innovator, which positions it as a potential rival to Nvidia as it could reduce dependency on the company’s chips, something America seems to use as a bargaining chip now and then.
The VLA model will work together with Xpeng’s VLM (Vision-Language Model) system, enabling ultra-low-latency interaction between the driver, vehicle, and autonomous system. The chip will be manufactured with a 3-4 nanometer processing node, the China CEO of Volkswagen’s subsidiary Cariad, Frank Han, has said, with “few options” for foundries capable of mass producing it.
According to him, Volkswagen currently plans to deploy the chip in its third generation of China Electrical Architecture (CEA). By 2030, 80% of Volkswagen Group cars sold in China are expected to be developed with CEA. Han also claims that the first model of the architecture powered by Horizon Robotics J-series chips will be rolled out at the end of this year.
Claim your free seat in an exclusive crypto trading community - limited to 1,000 members.
You May Also Like
Google's AP2 protocol has been released. Does encrypted AI still have a chance?
GitHub Copilot Gets Smarter With Context Engineering Techniques
