Caterpillar is expanding its integration of artificial intelligence and automation across its construction equipment lineup by teaming up with chipmaker Nvidia. The heavy machinery leader is currently testing an AI-assisted system in its mid-sized Cat 306 CR Mini Excavator. Known as “Cat AI,” this technology was developed using Nvidia’s Jetson Thor physical AI platform and is being showcased at CES this Wednesday.

A major advantage of implementing such systems is the valuable operational data they gather and transmit. “Our customers don’t spend their days in front of a laptop—they’re out working on-site,” Hootman explained. “Having access to real-time insights and being able to act on them while they’re on the job is extremely important to them.”

In addition, Caterpillar is experimenting with digital replicas of construction sites through Nvidia’s Omniverse simulation tools. These digital twins help evaluate scheduling options and improve estimates for required building materials. Hootman noted that Caterpillar’s equipment sends approximately 2,000 data messages to the company each second, information that will support these simulation efforts.

While Caterpillar already operates fully autonomous vehicles in mining, these pilot programs represent a strategic move to introduce more automation across its broader equipment range. “We chose to start here because it addresses a current challenge our customers face, and it’s an area where we already had strong momentum and could move to market relatively quickly,” Hootman said. “It also establishes a technological foundation we can build on in the future.”

Collaborating with established industrial brands like Caterpillar aligns well with Nvidia’s focus on physical AI systems. During its CES keynote on Monday, Nvidia outlined its comprehensive ecosystem for physical AI, which includes open AI models such as its Cosmos family, simulation resources, and developer kits. “Physical AI represents the next evolution of artificial intelligence,” Talla stated. “Nvidia is leading this shift by providing the computing power to train models, simulate their performance, and deploy them into robots—whether in autonomous vehicles or Caterpillar machinery.”