Meta is experimenting with robots that can perform routine physical tasks inside data centres, including moving equipment, pressing controls, inspecting racks and handling some cable-related work.
The trials reportedly involve robotic arms and mobile platforms from several hardware suppliers. The goal is not yet a fully autonomous data centre. Current systems remain limited by perception, dexterity, battery life, mobility and the difficulty of operating safely in spaces designed for human technicians.
Why data centres are attractive environments for robotics
Compared with homes or public streets, data centres are structured and repeatable. Server racks follow planned layouts, access is controlled and many maintenance tasks can be documented precisely. Those characteristics make the environment suitable for incremental automation.
The business pressure is also strong. Hyperscale facilities operate continuously, downtime is costly and AI expansion is creating more equipment to install and maintain. Robots could handle repetitive inspections, transport heavy components or enter hazardous areas while skilled engineers focus on diagnosis and complex repairs.
The work robots still struggle to perform
- Manipulating flexible cables without damaging connectors.
- Navigating crowded aisles while people are working.
- Recognizing unusual configurations that differ from digital plans.
- Operating for long periods without recharging.
- Recovering safely when a task fails halfway through.
These limitations matter because a data centre is not a demonstration laboratory. A robot that succeeds 95 percent of the time may still create unacceptable risk if the remaining failures disconnect critical systems or block emergency access.
The workforce question
Automation may reduce demand for some repetitive roles while increasing demand for robotics technicians, safety specialists, remote supervisors and systems integrators. The transition will depend on whether companies use robotics mainly to increase capacity or primarily to reduce labour costs.
Communities also need more precise employment projections. Large technology facilities often receive public incentives based partly on job creation. If automation changes the permanent workforce, governments should revisit whether incentives still reflect the actual long-term public benefit.
MaryChuks analysis
The most credible design is human-supervised robotics rather than the fantasy of an empty facility managed entirely by machines. Robots can extend reach and consistency, but humans remain necessary for judgement, accountability and response to novel failures.
For future offshore research habitats and data grids, this model becomes even more important. Robots could inspect submerged energy equipment, corrosive exterior structures and restricted server zones, while trained human teams retain command authority. Physical AI is most valuable when it strengthens resilience without concealing who remains responsible.
Source: Wired.
Discover more from Marychuks.com AI, Psychology, Business & CreativeVerse
Subscribe to get the latest posts sent to your email.
