Kodiak AI picks AWS as primary cloud provider to power autonomous trucking AI
Kodiak AI is using Amazon Web Services as its primary cloud provider to run the compute-heavy AI development and safety simulations behind its planned driverless highway launch.
- later in 2026
- minutes
- tens of thousands of miles
- millions
What Happened
Kodiak AI, Inc. (Nasdaq: KDK) announced it is using Amazon Web Services (AWS) as its primary cloud provider to power the compute-intensive AI development and safety simulations critical to autonomous trucking technology. The Mountain View, California-based company says the infrastructure matters as it prepares to launch driverless service on public highways later this year. Kodiak describes itself as a provider of Physical AI-powered autonomous driving technology.
- BreakPoint, Kodiak's in-house, AI-powered safety validation tool
- Cloud-based simulation
- AI model development
- GPU compute capacity to refine the AI models that power Kodiak Driver, the company's autonomous driving system
- Customer-facing operations center software
- Storage of data critical to operations
minutes
BreakPoint identifies potential edge cases in minutes that might otherwise take tens of thousands of miles of real-world driving to encounter.
“AWS cloud services enhance Kodiak's ability to launch and scale our driverless operations safely and reliably. AWS is the proven leader in cloud compute and storage, and exactly the partner we need to scale our in-house simulation program, train our AI models, run our customer-facing ops center software, and store data critical to our operations.”
Kodiak says BreakPoint deliberately injects realistic, time-varying errors into the autonomy system's signals and adversarially searches for the rare edge cases that could lead to a collision. AWS also provides the GPU compute capacity Kodiak needs to refine the AI models powering Kodiak Driver. As Kodiak readies its launch of driverless trucks on public highways later in 2026 in Texas, the company says the collaboration gives it the cloud infrastructure needed to support future expansion at scale across the United States.
“Getting autonomous driving right demands enormous compute power — especially when it comes to safety validation. AWS gives Kodiak the scalable cloud infrastructure to run millions of simulated driving scenarios, train AI models, and store the critical data behind real-world autonomous operations. As Kodiak moves toward driverless service on public highways, AWS provides the backbone to do it safely and at scale.”
Previously from Kodiak Robotics, Inc.
Kodiak has been working toward unsupervised highway operations in Texas. In September 2026, our coverage reported that Kodiak AI named the 219-mile Dallas-Houston lane as its first driverless long-haul route, targeting a launch by the end of 2026 after months of intervention-free test deliveries with no human at the wheel. We also reported that IKEA will become Kodiak's first driverless launch shipper partner, with freight runs between the Houston and Dallas areas set to begin later in 2026.
- Kodiak AI names Dallas-Houston as first driverless long-haul lane, targets year-end launch
- Kodiak and IKEA to launch driverless truck service on Texas I-45 route in 2026
Background drawn from MotorClaw's earlier coverage of Kodiak Robotics, Inc.'s official releases.
Why this matters
Autonomous trucking depends on enormous computing power to validate safety and train driving models before the human in the cab is removed. This deal gives Kodiak cloud capacity it says it needs to prepare for driverless service on public highways later in 2026, starting in Texas, and to support expansion across the United States.
Terms in This Story
- Physical AI
- Artificial intelligence that operates in the physical world, such as software controlling machines and vehicles, rather than only in digital settings.
- Edge case
- A rare or unusual situation that a system may not handle correctly, and that is hard to encounter during ordinary testing.
- GPU compute
- Processing power from graphics processing units, which are used to handle the heavy parallel calculations required to train AI models and run simulations.
- Simulation
- Software-based recreations of driving situations used to test and validate how an automated system behaves without putting vehicles on the road.
Summarised from the linked release; details can be imperfect — always verify against the original source.