Summary
- Exclaim Robotics has raised $4.95m in pre-seed capital to build its first data-centre maintenance prototypes.
- The proposed robots would handle precise rack-side tasks including optic replacement, cable reconnection, and drive reseating.
- Commercial deployment depends on proving repeatability, safety, dexterity, and integration with tightly controlled operating procedures.
Exclaim Robotics has raised $4.95m in pre-seed funding to develop autonomous mobile robots for routine rack-side maintenance inside AI data centres.
The Zurich-based startup intends to build a remote-hands automation platform capable of navigating a data hall, positioning itself in front of a rack, and carrying out precise tasks from floor level to the top rail without alterations to the building.
Proposed work includes cleaning and replacing optical components, reconnecting copper cabling, and reseating drives. The company is concentrating on dexterous manipulation rather than the movement of heavy equipment.
Founderful and Playfair Capital co-led the funding round. The capital will be used to build initial prototypes, purchase robotics hardware, and expand the engineering team before the company attempts live deployments.
Automation reaches the rack face
Data centre automation is already extensive in monitoring, workload management, environmental control, and fault detection. Physical intervention at the rack remains far more dependent on technicians because equipment layouts vary and mistakes can interrupt live services.
A robot operating in that environment must do more than identify the correct cabinet. It needs to distinguish ports and components, handle small tolerances, manage flexible cables, avoid neighbouring equipment, and verify that each action has been completed correctly.
Those requirements become harder where racks contain mixed generations of hardware, undocumented changes, damaged labels, or cabling installed differently from the original design. A technician can often interpret such irregularities from context; an autonomous system must either recognise them reliably or stop and escalate the task.
Exclaim’s founder, Helen Oleynikova, has linked the company’s proposition to the prospect of higher-voltage direct-current systems entering future AI rack designs, where routine intervention could require stricter controls and specialised procedures.
The voltage transition is not yet universal, and any robot would still need to operate across conventional alternating-current and lower-voltage direct-current environments. Its safety case would have to cover electrical risks, moving mechanisms, emergency stops, access control, fire systems, and interaction with technicians.
Prototype capital, not an operating fleet
The $4.95m round finances an engineering programme rather than a commercial deployment. Exclaim’s first year is expected to focus on proving a limited set of tasks repeatedly without damaging equipment.
That is an appropriate test for critical infrastructure. A rack-maintenance robot does not need to fail often for the operational consequences to outweigh its labour savings. Disconnecting the wrong cable, damaging an optic, or obstructing an aisle could affect availability and invalidate change-control procedures.
Operators would also need to integrate the platform with asset records, ticketing systems, security permissions, camera retention, and maintenance workflows. Each action would require a traceable instruction, verification step, and rollback or escalation path.
The business case may be strongest where sites are remote, staffing is scarce, or simple interventions cause long delays because an authorised technician has to travel. Robots could also handle repetitive inspection and low-level maintenance while keeping people away from energised equipment.
They are less likely to replace experienced remote-hands teams across the full range of faults. Data centres contain mechanical, electrical, network, and server problems that require diagnosis, judgement, and work outside the rack face.
Oleynikova’s previous work includes roles at Google, Microsoft, Nvidia, and ETH Zurich. The company has added engineers with experience from Nvidia and ETH as it develops the first system.
The startup’s next milestone is not the funding round but repeatable operation in a representative data hall. Only after the robot can identify, manipulate, verify, and document components without disrupting adjacent services will the platform move from an attractive labour thesis to usable operational infrastructure.
