Kodiak AI, a supplier of bodily AI with a concentrate on AI-powered autonomous automobile expertise, has introduced Nvidia expertise will assist energy Kodiak’s next-generation autonomous-driving answer.
Kodiak intends to speed up the deployment and scaling of driverless autos with the Nvidia Drive Hyperion architecture, a production-ready reference platform that’s underpinned by two Nvidia Drive AGX Thor centralized computer systems.
Constructed on the Blackwell structure, Drive AGX Thor delivers as much as 1,000 INT8 TOPS / 2,000 FP4 TFLOPS of AI compute. This large compute headroom is optimized for the newest generative AI, transformer, and Imaginative and prescient Language Fashions (VLM), enabling the Kodiak Driver to interpret nuanced and unpredictable real-world circumstances in actual time to allow secure and verifiable driving habits.
“The important thing to deploying Bodily AI is harnessing the facility of datacenter-scale AI and optimizing it to run on the edge,” stated Don Burnette, founder and CEO of Kodiak. “At a time when our core AI capabilities and product footprint are quickly increasing, Nvidia Drive Hyperion delivers strong efficiency throughout a variety of platforms for assembly the problem of scaling autonomous autos.”
Unmatched real-world expertise
Kodiak enters this collaboration as a longtime autonomous trucking chief. With the Kodiak Driver deployed in 20 vehicles as of the tip of 2025, Kodiak is the one firm offering driverless service on customer-owned Class 8 vehicles with no people within the cab right now.
Kodiak-powered vehicles function 24/7 in West Texas’s harsh and demanding Permian Basin, navigating bumps, potholes, cattle guards, oncoming autos, and mud storms and not using a human on board.
