Washington | 26°C (scattered clouds)
Meta Turns to Robots for Data‑Center Upkeep

Meta experiments with robot arms to automate data‑center maintenance, signaling a shift away from human crews

Meta is testing robotic systems from ABB, Kinova and Watney Robotics to handle routine tasks in its data centers, potentially cutting human labor by up to 80 %.

Since the data‑center boom first promised a surge of local jobs, the reality has always been a bit… overstated. Most of the work is short‑term construction, and the ongoing crew that monitors racks is lean at best.

Now that lean crew might be disappearing altogether. Recent reporting from Ars Technica reveals that Meta has begun trial‑running a handful of robot arms inside its facilities. The hardware comes from several vendors—ABB, Kinova and Watney Robotics—each offering a different style of manipulator for tasks like swapping network cables, resetting power switches and checking temperature sensors.

One Meta employee, speaking on the condition of anonymity, said the Kinova arm alone could take over roughly 80 percent of the manual workload. “We thought those of us performing the physical tasks were safe for a while, but not anymore,” the staffer admitted, a hint of resignation in the voice.

The numbers line up with the economics of a data center. The biggest expense is the silicon—expensive AI chips that power large language models—not the people who walk the aisles. Even when you factor in the initial build‑out, human labor is a drop in the bucket, so Meta’s push to automate maintenance is really about shaving costs wherever possible.

Meta declined to comment when approached for a statement about the robot trials. If the tests prove successful, we may soon see racks serviced by silent mechanical arms instead of the occasional human technician, marking another step toward a fully autonomous AI infrastructure.

Comments 0
Please login to post a comment. Login
No approved comments yet.

Editorial note: Nishadil may use AI assistance for news drafting and formatting. Readers can report issues from this page, and material corrections are reviewed under our editorial standards.