Exclaim Robotics raises €4.29M Pre-Seed to send robots inside AI data centres before 800V kills someone

Exclaim Robotics (Zürich) has raised €4.29 million ($4.95 million) in a Pre-Seed round co-led by Founderful and Playfair, exiting stealth to build autonomous mobile robots for AI data centre maintenance. The company was founded in 2026 by Helen Oleynikova, who spent 15 years across Willow Garage, Google, Microsoft, NVIDIA, and ETH Zürich, holds a PhD from ETH Zürich in mapping and planning for cluttered, unstructured environments, and co-created nvblox — the 3D mapping and collision-avoidance library now shipped in NVIDIA Isaac ROS. Founding engineer Patrick Pfreundschuh, who created the BIEVR-LIO LiDAR-inertial odometry system and was part of the winning team at the DARPA Subterranean Challenge, joins from the same ecosystem.

The problem: 800 volts at the rack face

Almost everything in a modern data centre is managed remotely: provisioning, scheduling, failover, networking. The physical layer is the exception. When hardware fails, someone has to walk to the rack.

The next generation of AI data centres will distribute power at 800 volts DC directly inside the rack. At that voltage, opening a rack carries arc-flash risk, and ordinary break-fix work — replacing a failed drive, reseating an optical transceiver — becomes genuinely hazardous. The shift comes against a backdrop of AI data centre buildout that is outpacing the supply of trained technicians: the human bottleneck and the safety hazard are arriving together.

Exclaim is starting with the highest-volume break-fix tasks: replacing and cleaning optics and SFPs, copper network cabling, and drive replacement. The robot navigates the data hall autonomously, positions itself at a rack, and works the front face from floor to top rail without modifications to the building.

The approach: simulation first, human hand-off by exception

Exclaim's training methodology begins in simulation, where a robot can practise a task far more times than is possible on real hardware, then transfers to real racks until the system is reliable enough for production. The harder problem, the company acknowledges, is environmental rather than mechanical: racks are dense with cabling, and a robot has to understand what it is looking at before it can safely touch anything. Where confidence is low, it hands off to a remote operator rather than attempting the task — a design choice that puts judgment before throughput.

The longer-term target is any infrastructure where maintenance is hazardous, remote, or hard to reach. High-voltage and energy infrastructure is named as the next vertical after data centres.

"AI data centres are scaling faster than anyone can staff or safely maintain them, and the 800V transition will change everything about how that infrastructure gets run," said Antonia Albert, Principal at Founderful. "Helen and her team from ETH Zürich, NVIDIA, Microsoft, and Google are the people to solve it. We backed Exclaim because Helen is exactly the founder you want when a hard problem becomes urgent."

The Pre-Seed funds Exclaim's first prototypes, hardware acquisition, and engineering team growth in Zürich. Whether the company can compress the build-and-validate cycle fast enough to land a named operator endorsement before the competitive window tightens is the question this round opens.

A thread from the archive: Playfair previously backed Astral Systems — a Bristol robotics company that raised £23 million in Series A earlier this year. In Exclaim, the fund moves from precision-manufacturing robotics to infrastructure-maintenance robotics: a consistent thesis in autonomous physical systems, at different points of the maturity curve.