Prompt and Model
Building with AI

Meta Tests Data Center Robots

Meta is testing robots from vendors like Watney, Kinova, and ABB to perform tasks such as plugging cables and resetting servers in its data centers, aiming

Meta is testing robots from vendors like Watney, Kinova, and ABB to perform tasks such as plugging cables and resetting...

Meta is testing robots to plug in cables, reset servers, and handle other tasks inside its data centers, according to several current and former workers. The effort, which may let Meta operate its expanding data centers with fewer people, involves hardware from vendors including Watney Robotics, Kinova, and ABB.

Company spokesperson Francis Brennan stated Meta is investing heavily in training and hiring workers, citing a major shortage of skilled labor. However, senior robotics manager Eric Xu said last year that the company's long-term goals are to put robots in data centers to speed up incident response and handle maintenance.

Current Experiments

In one experiment, Meta is evaluating a Kinova Gen3 robotic arm for power cycling servers. The company is also testing a different robot to swap networking cables. One data center worker estimates successful automation could replace up to 80 percent of some people's workloads. A relatively simple robot, described as looking like a finger or a pointy stick, is already in use at some facilities to press power buttons on devices like Mac Minis when remotely prompted.

Until recently, robots were often expensive and poorly suited for delicate data center work, sometimes crushing servers in earlier industry trials. The hardware has gotten cheaper, and the AI models powering robots are now far more capable. Several startups are working to build commercial robots for assembling servers and replacing wires.

Worker Concerns and Training

In Altoona, Iowa, home to Meta's largest data center campus, worker group chats have taken on a demoralized tone about being replaced by robots. Two workers say Meta is developing AI software and other technologies they believe will allow it to hire lower-skilled, lower-paid workers and centralize roles in areas like Denver, Colorado. One worker contends the company now wants 'smart hands' who just follow AI instructions.

Meta contends its need for skilled workers remains robust. This year, the company launched a program to train thousands annually in electrical, mechanical, and plumbing work at no cost, guaranteeing employment in states like Louisiana and Texas for those who complete it. The company also partnered with building trade unions to fund apprenticeships.

Altoona Mayor Dean O'Connor says he hasn't given much thought to robots taking jobs, framing automation as inevitable. The city has viewed tax breaks for Meta's data center, which save the company tens of millions annually, as a worthwhile trade-off for the jobs.

Deployed and Tested Robots

Meta has previously touted using a self-driving tugger robot to transport server racks and a wheeled, barcode-reading inventory robot developed in-house. These robots are now operating in data centers in Iowa and Virginia. The inventory bot is also used for failure inspections but cannot differentiate between green and red indicator lights because its camera only detects grayscale. It struggles with corners and cables, and requires a human to move it between buildings.

Two sources describe the inventory bot as too slow. Meta previously explored using drones as a faster alternative but did not pursue the idea.

The Altoona campus is often where Meta prototypes new technologies. Since last June, Meta has been trying out a pair of dual-armed Watney robots for cabling work there, supervised by humans. Watney, which launched in 2023, pivoted to data center robots early last year.

Meta is testing additional robots in New Albany, Ohio, at its newest Prometheus campus. Four-wheel robots with a scissor-lift-style riser and a six-axis arm on top, made by ABB, are being used to reseat parts and may eventually take on other tasks. The idea that robots will prevail over humans is not guaranteed, with one source noting many pilots but no provable working systems at scale yet.

Related coverage

More from Building with AI