Robots are moving into Meta’s data centers, and they are not arriving for demonstrations. They are being tested on the physical work that keeps servers running: plugging in cables, resetting machines, swapping network connections, and cutting power when needed.
The goal reaches beyond a few automated tools. Meta’s ongoing effort may allow the company to operate data centers with fewer humans, reducing labor and costs while changing the jobs performed inside these massive facilities.
Robots Take On the Work Behind the Servers
Meta is testing robots from several vendors, including Watney Robotics, Kinova, and ABB. The machines are aimed at tasks that technicians handle today, giving the company a way to test whether robots can work inside server-filled environments without damaging equipment or slowing operations.
One experiment uses a Kinova Gen3 robotic arm for power cycling, which means cutting off electricity to servers and restoring it. Meta is also testing a robot that can swap networking cables, a task that demands careful movement around dense hardware and connections.
Another test takes a simpler approach. Meta adopted a robot that presses the power button on devices to restart them, showing how automation can begin with a small, focused job instead of a machine designed to handle every task in a data center.
A Meta data center worker estimates that, if the cable-swapping robot succeeds, it could replace up to 80 percent of some workers’ workloads. That estimate points to a major shift in how data center labor could be organized, even if the robots begin with only a narrow set of duties.
The change is already raising concern among workers. Meta worker group chats have expressed fear that robots will lead to job losses, and one Meta data center worker said, “We thought those of us performing the physical tasks were safe for a while, but not anymore.”
Why Meta Sees a Bigger Role for Machines
Eric Xu, senior manager for robotics at Meta, said, “Long-term goals are to put robots in data centers, arguing they could speed up incident response, monitor the environment, and handle preventative maintenance.” Those tasks would place robots at the center of daily operations, not just in controlled trials.
Meta’s push comes as the technology behind these machines changes. Robots were previously expensive and sometimes damaged servers during trials, but hardware costs have decreased and AI models are now more capable. Those changes give Meta more room to test machines in real facilities and measure what they can do.
Several startups are also working on commercial robots capable of assembling servers, swapping parts, and replacing wires. Together, these efforts are aimed at the physical layer of computing, where a server still needs a cable, a replacement part, or a power reset even when software handles much of the surrounding work.
Supporters see a practical benefit: robots could take on repetitive and dangerous tasks, freeing workers to focus on more complex challenges. Robots could also make data centers in inhospitable environments more feasible, including facilities located underwater or in space.
Paul Golding, who oversees physical intelligence at Analog Devices, said, “More and more of our customers are telling us ‘we want to run our data centers using humanoid [robots]’ because you can run them at higher temperatures, in the dark, and all these things.” The statement captures the appeal of machines that can work where human access is difficult.
A Labor Experiment Inside a Growing Infrastructure Business
Meta is investing heavily in training and hiring workers to build and operate its data centers, creating a tension at the heart of the robotics effort. The company is expanding the human workforce needed to support its infrastructure while testing machines that could reduce the number of people required for some physical tasks.
Meta operates its largest data center campus in Altoona, Iowa, where hundreds of workers are employed. Meta opened the facility in 2014 and has a property tax break that saves tens of millions of dollars annually, making the campus an important part of the company’s infrastructure operations.
That workforce may face a new model as Meta develops AI software and technologies that could allow it to hire lower-skilled and lower-paid workers while centralizing roles in more advantageous locations. Robotics could support that model by handling physical duties that once required more specialized technicians on site.
The Altoona campus and other data centers now represent more than computing capacity. They are becoming test sites for a new relationship between people, AI software, and machines that can manipulate the hardware beneath digital services.
Meta’s experiments remain focused on cables, power buttons, server resets, and related tasks, but the direction is clear. As robots become less expensive and AI models become more capable, the machines operating inside data centers could move from isolated trials toward a permanent role in how Meta builds, maintains, and expands its infrastructure.
Based on
