AMD lands Kodiak, a first in autonomous truck compute

AMD EPYC processors will power Kodiak AI’s seventh-generation autonomous truck platform, the two companies announced Wednesday. Kodiak (Nasdaq: KDK) is the first autonomous trucking company to deploy the advanced EPYC chips in its hardware. EPYC chips are server processors, the same class of CPUs that run cloud and data-center workloads.

The truck’s job for those chips is autonomous truck compute: ingest a firehose of sensor data, clean it up fast and hand it to the software. Cameras, lidars and radars in Kodiak’s SensorPods generate that stream every second. The EPYC processors aggregate and preprocess it before the Kodiak Driver acts.

Kodiak already runs trucks with no one in the cab in the Permian Basin. The company’s second-quarter 2026 results showed seven additional driverless trucks deployed in the quarter, 35 customer-owned vehicles at quarter-end and more than 40,000 cumulative hours of paid driverless operation through Q2.

Driverless operations on public highways are targeted for later this year. The compute upgrade lands ahead of that launch. An autonomous truck’s need for compute speed The choice behind the compute provider centers on I/O and latency. Sensor processing, localization and path planning are latency-sensitive and do not split cleanly across hundreds of cores.

Clock speed and PCIe lanes are more important than a headline core count. EPYC is built for that kind of data movement. EPYC gives the Kodiak platform 80 PCIe lanes to move data at high rates. Base frequency is 3. 15 GHz. Max boost reaches 4. 4 GHz. AMD estimates a 25% clock-speed gain over the previous generation, at lower power consumption.

“Physical AI requires compute capabilities that can rapidly handle enormous amounts of sensor data while supporting increasingly sophisticated AI workloads,” said Don Burnette, founder and CEO of Kodiak. “AMD delivers best-in-class CPU technology with the performance, scalability and efficiency required for advanced autonomous driving applications.

AMD EPYC CPUs give us the compute capabilities we need to advance the Kodiak Driver and continue commercializing autonomous trucking.” Off-the-shelf silicon ahead of the highway launch Kodiak is using commercially available hardware from AMD, keeping its engineering resources on scalable, production-ready trucks.

That approach matters more as the driverless fleet in the Permian Basin grows. The Kodiak Driver runs today in long-haul trucking, industrial trucking and defense. Commercial partners include Atlas Energy Solutions, IKEA, J. B. Hunt, Bridgestone, Werner Enterprises, C. R. England, General Dynamics Land Systems and Roehl Transport.

Kodiak sells the system through a Driver-as-a-Service model and strategic partnerships. “Physical AI is pushing computing into some of the most complex and data-intensive environments in the world,” said Wayne Lyons, senior director of the automotive market at AMD.

“Kodiak is demonstrating what’s possible when advanced autonomous driving software is paired with high-performance AMD compute. Together with Kodiak, AMD is providing the processing capabilities needed to move driverless trucking from innovation to significant scale.”

The post AMD lands Kodiak, a first in autonomous truck compute appeared first on FreightWaves.