Torc Robotics Inc. has announced a major collaboration with NVIDIA Corp. and Flextronics International Ltd. to create a scalable, high-performance physical artificial intelligence (AI) compute system designed specifically for autonomous trucks. The announcement was made during Torc’s debut at NVIDIA’s GTC event in San Jose, California, where the company showcased its latest advancements in autonomous driving technology.

“This was a thrilling week for us,” said CJ King, chief technology officer at Torc, during the event. “Partnering with NVIDIA and Flex is a key milestone as we edge closer to the commercialization of our autonomous trucking solution. GTC gave us the opportunity to showcase our technology and introduce our engineering talent to the broader tech and automotive communities.”

Founded in 2005 and now an independent subsidiary of Daimler Truck AG, Torc Robotics is focused on developing self-driving software for commercial long-haul trucking in the U.S. The company’s latest efforts are centered around what it calls “physical AI” — a system that enables trucks to operate autonomously in the real world by processing complex environmental data through sensors like lidar, radar, and cameras. Torc claims this is the first physical AI system deployed for autonomous long-haul trucking at production scale.

With headquarters in Blacksburg, Virginia, and additional engineering offices in Austin, Texas, Montreal, Quebec, and operations in Ann Arbor, Michigan, and Dallas-Fort Worth, Texas, Torc is positioning itself as a national leader in autonomous freight solutions.

At the core of this new compute platform is NVIDIA’s DRIVE AGX system powered by the DRIVE Orin system-on-a-chip (SoC) and the DriveOS operating system. According to NVIDIA’s Vice President of Automotive, Rishi Dhall, “DRIVE AGX is already proven in full-scale automotive production and offers the performance and low-latency capabilities needed for Torc’s demanding AI software stack. It’s designed to handle the critical real-time requirements for safe, reliable autonomous operation.”

Torc is also preparing to upgrade to NVIDIA’s next-generation DRIVE Thor platform as part of its roadmap toward commercial launch in 2027.

Flextronics is contributing its Jupiter compute design platform and advanced manufacturing capabilities, helping Torc build a scalable production system based on an autonomous-ready Freightliner Cascadia truck. This vehicle is equipped with redundant systems and advanced technologies tailored for autonomous driving deployment.

“Our partnership with Torc, Daimler Truck, and NVIDIA reflects how Flex works across the full automotive ecosystem to enable next-generation mobility solutions,” said Mike Thoeny, president of automotive at Flex. “We’re proud to help bring autonomous long-haul trucking closer to reality through our Jupiter platform and production expertise.”

Torc emphasized that this compute system is engineered to meet strict size, cost, performance, and reliability standards — aligning with the total cost of ownership (TCO) goals of its fleet customers. The company also highlighted the adaptability of the platform, calling it a true software-defined vehicle (SDV) capable of adjusting to new operational design domains, from changing road conditions to updated sensor configurations and routing.

Last year, Torc successfully validated its driverless platform on a closed course at highway speeds, marking a major step forward in readiness for real-world deployment.

As the autonomous trucking industry races toward commercialization, Torc’s collaboration with NVIDIA and Flex represents a significant leap forward in building the robust, scalable infrastructure required to bring driverless long-haul freight transport to U.S. highways.

By Impact Lab