NEURA Mobile Robots is acquiring 100 percent of the shares in ADLATUS Robotics GmbH, an established provider of autonomous cleaning robotics with more than 20 years of experience. With this acquisition, several hundred installed systems, proprietary navigation software, and application expertise in the industrial, logistics, healthcare, and public building sectors are being transferred to the NEURA Group.

Company founder David Reger articulates NEURA’s plans with unusual candor: The company is not acquiring ADLATUS to add another cleaning robot to its portfolio, but to give such machines a new “brain.” Cleaning, he says, is a massive global labor market and, at the same time, a prime example of what Physical AI can transform. In the future, the systems are to be equipped with additional sensors and gradually connected to the Neuraverse—evolving from fixed route navigation to recognizing surfaces and types of dirt, selecting an appropriate strategy, and learning from ongoing operations.

Bosch Rexroth Sells a Platform

The acquisition made eleven days earlier follows the same logic. Effective October 1, 2026, NEURA will take over Bosch Rexroth’s ACTIVE Shuttle autonomous guided vehicle in its entirety—including hardware, software, and, in the future, service as well. The proven ACTIVE Fleet Manager technology and ROKIT navigation software will remain part of the solution and, in the long term, will also be made available as tools for other NEURA applications via the Neuraverse.

For existing customers, this is primarily a matter of continuity. Both sides emphasize that they have prepared for the handover together; service, support, and investment security are the top priorities. The direction of the deal is noteworthy: A traditional German mechanical engineering conglomerate is divesting itself of an established AMR platform, while a ten-year-old AI company is stepping in to acquire it.

The Bet: Platform Beats Product

Behind this lies an argument that extends beyond NEURA. Unlike generative AI, Physical AI thrives on real-world data and experience—machines must learn how physical tasks are actually performed. ADLATUS’s installed base provides the foundation for exactly that: developing, validating, and scaling under real-world conditions. Anyone who wants to train models needs fleets in the field. Acquisitions are the fastest way to get there.

The foundation for this was laid in June. The Series C funding round, totaling up to $1.4 billion, brought together Tether, Qualcomm, Amazon, and Nvidia, along with Bosch, Schaeffler, and the European Investment Bank; the company’s valuation stood at around $7 billion. NEURA Mobile Robots operates under the established brand ek robotics, is headquartered in Hamburg, maintains locations in Reutlingen, Milan, Prague, and Buckingham, and employs more than 300 people.

Open Ecosystem—or Dependency?

NEURA emphasizes that the Neuraverse is intended to be an open infrastructure for different robots, manufacturers, and designs; even competitors should be able to connect their machines. According to Reger, Physical AI can only scale as an ecosystem.

For users, this is the crucial question to consider. Anyone currently operating ACTIVE Shuttles in material flow should clarify three points in writing: service commitments beyond October 1, the roadmap for firmware and interfaces—and, above all, whether fleet management and navigation data will remain exportable should the platform strategy take a different direction. An open ecosystem is not recognized by the announcement, but by the exit clause.