Caterpillar partners with Nvidia to integrate AI into its construction machinery, piloting the Cat AI assistive system on the Cat 306 CR Mini Excavator using Nvidia’s Jetson Thor platform, with a demonstration at CES on Wednesday.
The Cat AI system operates on a fleet of AI agents. Brandon Hootman, vice president of data and AI at Caterpillar, explained its functions to TechCrunch. The system answers questions from machine operators. It provides access to necessary resources. Operators receive safety tips through the interface. The technology also enables scheduling of services directly from the machine. This integration occurs within the mid-size Cat 306 CR Mini Excavator, a model commonly used in construction tasks requiring precision and maneuverability in tight spaces.
A primary advantage lies in the data collection capabilities of the Cat AI system. Machines equipped with this technology gather operational data and transmit it back to Caterpillar. This process delivers real-time insights to operators active in field environments. Hootman emphasized the practical context of operators’ work conditions. He stated, “Our customers don’t live in front of a laptop day in and day out; they live in the dirt.” He continued, “The ability to get the insights and take the action that they need while they’re doing the work is very important to them.” Such data flow supports immediate decision-making during construction activities without requiring operators to leave their equipment.
Caterpillar extends its AI initiatives through pilots of digital twins for construction sites. These digital replicas utilize Nvidia’s Omniverse library of simulation resources. The simulations test various scheduling scenarios for projects. They also calculate precise requirements for building materials. Hootman noted the volume of data generated by Caterpillar’s machines. Each machine sends roughly 2,000 messages per second to the company. This continuous stream of data populates and refines the digital twin simulations, enhancing accuracy in project planning and resource allocation.
Caterpillar maintains fully autonomous vehicles within its mining sector operations. Hootman described the current pilots as a progression in automation efforts. These programs advance the integration of autonomous features across Caterpillar’s broader equipment portfolio, building on established mining applications to construction machinery.
Nvidia positions physical AI as central to its strategy, aligning with partnerships like the one with Caterpillar. Bill Dally, Nvidia’s chief scientist, addressed this focus in a 2025 TechCrunch interview. He stated that the chipmaker considers physical AI the next frontier for the company and its powerful GPUs. This direction emphasizes applications beyond traditional computing into machinery and robotics.
During Nvidia’s CES keynote on Monday, the company detailed its full-stack ecosystem for physical AI. Components include open AI models from the Cosmos model family. The ecosystem incorporates simulation tools for model development and testing. It provides developer kits to facilitate deployment. These elements support comprehensive physical AI implementation across industries.
Deepu Talla, Nvidia’s vice president of robotics and edge AI, broadened the scope of physical AI in discussions with TechCrunch. He rejected a narrow view limited to robotics companies. Talla asserted, “Physical AI is the next wave of AI.” He elaborated, “Nvidia is pioneering that with computers that train the models, that do the simulation to test the models and deploy the models into the robots, whether that’s an autonomous car or a Caterpillar machine.” This approach encompasses diverse applications, from vehicles to heavy construction equipment.
