German robotics startup Neura Robotics has partnered with Germany’s RWTH Aachen University to establish a robotics training facility designed to supply real-world training data for robotic systems.

Plans for the new site were unveiled last week. The facility will become part of Neura’s existing network of physical AI training centers, which the company calls Neura Gyms. Neura said the centers are intended to address what it describes as the largest obstacle to deploying intelligent robots: the shortage of real-world training data. In robotics, that data is harder to gather than in many software-only AI systems because machines must be trained and validated against physical environments, tasks and safety requirements rather than only digital inputs.

The facilities combine physical training with high-fidelity simulation to create datasets that feed directly into Neuraverse, Neura’s open, cloud-based platform connecting robots, developers and industry partners.

Neura Gym rollout

Neura said the sites are also built to let partners train and validate robots for specific applications before moving to full industrial deployment, with the aim of reducing adoption risks and shortening deployment timelines.

The latest Neura Gym is scheduled to open by the end of 2026. Another 10 locations are under development, and Neura said half of them are expected to be operational by the end of the year across Europe, China and the U.S.

Alongside the RWTH Aachen University site, Neura’s second flagship facility is planned for Munich Airport in the fourth quarter. That site is being developed in collaboration with the Technical University of Munich and its Munich Institute of Robotics and Machine Intelligence.

Neura said the two centers together position Germany as the backbone of Europe’s physical AI infrastructure.

Phil Solis, an analyst at IDC, told AI Business that the launch at the German university site is notable because it is more than simply a place to test robots and collect data.

“It combines simulation, a place to physically test robots, and developers and other industry partners can all leverage this,” Solis said. “AI-based development of software is one step that is greatly accelerating improvements in robotics compared to traditional programming, and this accelerates the way various companies can work together.”

Solis said the model could also support what he called shared learning, in which improvements made in testing environments can be pushed out to robots operating on-site.

Neura’s global rollout is backed by the $1.4 billion the company raised last month in a Series C financing round.

Neura Gyms in a global context

Robotics gyms are already standard practice in China, where multiple regional hubs have been created to enable standardized and cooperative data collection, training and validation for physical AI systems, according to Himanshu Kumar Ojha, an analyst at Gartner.

“Compared to the digital world and knowledge tasks, availability and accessibility of relevant data is the limitation for physical AI,” Ojha said. “A scalable approach is necessary for collection and utilization for real-world data, to make these systems achieve necessary levels of performance for specific tasks and environments, through post-training and fine-tuning of base and foundation AI models.”

Ojha said Neura’s RWTH Aachen University center aligns with the recent scaling of physical AI in Europe as the region seeks to rival the more established robotics markets in the U.S. and China.

“This provides a collaboration platform for Germany and overall Europe’s existing ecosystem of manufacturing and technology companies to contribute to the advancement of physical AI, and benefit with its adoption by various industries,” he added.

Looking ahead, Ojha said the current lack of global standards for regulation and compliance in physical AI means robotics hardware manufacturers and systems integrators could use the platform for standardization efforts, validation and compliance testing.

“Such initiatives will accelerate regional innovation and reduce Europe’s dependence on China or U.S. for physical AI, which is increasingly becoming a strategically important technology domain,” he added.

Solis said the launch is focused more on data sovereignty than on global competition.

“China has the most deployed robots and thus has much more data that can be collected,” he said. “Neura Gyms’ data and intelligence layer will be in Germany. Companies will be able to access this from any country, but it will likely be countries like this that … hold most of the data.”