German robotics startup Neura Robotics has selected Italian firm SECO to engineer and build compute modules for its cognitive robots, in a partnership intended to industrialize and scale physical AI in Europe. The agreement brings together robot design, embedded computing and manufacturing capabilities, covering both the hardware needed inside the machines and applications for industrial environments.

The modules will integrate Qualcomm’s Dragonwing processors into Neura’s distributed compute architecture. Rather than relying on a single central computing system, the architecture distributes processing capabilities throughout the robots.

Neura calls this approach “Smart Limbs.” By placing compute closer to joints, sensors and other points of physical interaction, the system enables robots to process information locally and make decisions with lower latency. The arrangement also makes the compute modules a central part of how Neura’s robots interact with their physical surroundings, rather than treating processing as a separate, centralized layer.

The companies will initially develop modules for Neura’s humanoid robots, including its domestic robot 4NE1. They will also collaborate on physical AI applications for semiconductor and electronics manufacturing, with the aim of creating automation systems that can be applied globally. This gives the partnership both a consumer-robotics focus through 4NE1 and an industrial focus in sectors where Neura and SECO intend to develop additional applications.

“Physical AI will not be built by one company alone,” David Reger, founder and CEO of Neura, said in the release. “It requires an ecosystem across sensing, compute, cloud and industrialization … This is exactly how I believe Europe should approach Physical AI: combine our industrial knowledge, build together and scale it globally.”

The partnership is the latest development in Neura’s push into physical AI. The company completed a $1.4B funding round for the project in June, with investors including Nvidia, Amazon and Qualcomm participating.

In July, Neura opened several “robot gyms,” which it said were created to address the scarcity of real-world training data needed to develop and deploy intelligent robots. The facilities combine physical training with high-fidelity simulation to create datasets that feed directly into the Neuraverse, Neura’s platform for connecting robots, developers and industry partners. Together with the new compute-module work, the gyms provide a route from robot training and data generation to the hardware and applications used to deploy the systems.

As part of the agreement with SECO, the partners will establish a new Neura Gym in Italy. The facility will be Neura’s first in Southern Europe, adding a regional training site to the partnership’s compute-module and industrial-application efforts.

The partnership was announced by Neura Robotics in its official blog post:

September 9, 2026