Agility Robotics moves to NVIDIA hardware platform for Digit 5 as it targets commercial scale

Agility Robotics is expanding its collaboration with NVIDIA as it prepares Digit 5 for commercial scale, a step the companies say is aimed at standardizing the computing hardware inside the next generation of its humanoid robot. The update, reported by Yahoo Finance, centers on how Digit 5 will run AI workloads at the edge, without relying on continuous cloud connectivity.

The companies plan to consolidate Digit 5’s onboard compute using NVIDIA IGX Thor modules. In plain terms, IGX Thor is NVIDIA’s edge AI hardware platform designed to package processors and accelerators into modular compute blocks that system builders can integrate into robots, machines, and other devices that must make decisions locally.

Agility and NVIDIA also point to a longer-running relationship tied to humanoid safety. In the earlier phase of collaboration, the firms worked on safety-related capabilities for humanoid systems, and the Digit 5 announcement frames the expanded effort as a continuation of that work as the robot moves from development toward deployment.

Digit 5 is described in the report as a platform robot intended to be manufactured and deployed at larger scale than earlier generations. Consolidating onboard compute around a common hardware module is often used in robotics to reduce integration complexity, speed up iteration, and make it easier to scale production across multiple deployments, though the report does not provide further manufacturing or engineering metrics.

For NVIDIA, robotics is one of several growth areas that rely on AI compute deployed outside the data center. The company has positioned its edge platforms for real-world AI, spanning industries such as warehouse automation and industrial inspection, where latency and reliability matter. A partnership that ties a humanoid robot’s compute architecture to NVIDIA’s modular edge hardware fits that strategy, even as the report does not specify performance targets or model changes.

For Agility Robotics, the shift to a standardized compute stack is also likely to be tied to operational needs that emerge at scale, such as consistent software performance across fleets of robots and streamlined integration of perception, planning, and control. The report, however, does not disclose which specific AI models, sensor processing pipelines, or safety validation steps will run on the IGX Thor modules, nor does it provide any quantitative results.

What is still unclear from the reporting is the full technical scope of Digit 5’s system design beyond the compute modules, including the number of modules, expected power and thermal characteristics, and how software will be packaged for production. The announcement also does not detail timelines for first customer deployments or whether Agility intends to make the platform broadly compatible with third-party add-ons.