Meta Tests Robots for 80% of Some Data Center Tasks as AI Infrastructure Spending Soars
Updated
Updated · WIRED · Aug 28
Meta Tests Robots for 80% of Some Data Center Tasks as AI Infrastructure Spending Soars
3 articles · Updated · WIRED · Aug 28
Summary
Meta is quietly testing robots to swap network cables, reset servers and cut power in data centers, with one worker estimating a successful cabling bot could replace up to 80% of some employees’ workloads.
The push targets lower labor costs and faster operations as Meta rapidly expands AI infrastructure, though the company publicly says it still needs more skilled workers and is training thousands annually for data-center jobs.
In Altoona, Iowa, and New Albany, Ohio, Meta is evaluating machines from Watney, Kinova and ABB, while simpler button-pressing robots and earlier tugger and inventory bots are already operating at some sites.
The trials still face limits: robots can be slower than humans, need battery recharging, struggle with corners and cables, and are not yet suited for demanding work such as cabling Nvidia GB300 systems.
The experiments could complicate the jobs case Meta and other tech companies use to justify tax breaks, especially in communities like Altoona where the company saves tens of millions of dollars a year.
As Meta deploys clumsy robots to manage massive AI servers, will this automation solve infrastructure bottlenecks or trigger costly new hardware failures?
With rapid tent deployments and off-grid power fueling Meta's expansion, what hidden environmental costs lurk behind these automated data centers?