Meta Platforms is testing robots to perform hands-on maintenance inside its data centers, an automation effort that could eventually reduce the amount of human staff needed to run the company’s rapidly expanding AI data center buildout.

The company’s tests with robots, first reported in Wired, include robotic systems from Kinova, ABB, and Watney Robotics. The machines are being evaluated for jobs that include connecting network cables, restarting servers, reseating hardware and inspecting equipment. Some robots are already handling more basic duties, including transporting server racks and tracking inventory.

The technology remains far from fully autonomous. Robots under evaluation require human supervision and can be slower than technicians. Battery life is another limitation, while navigating crowded data center floors and working with tightly packed cables remain difficult tasks.

Meta has been testing two dual-arm robots from Watney for cabling work at its Altoona, Iowa, data center campus since 2025. At Meta’s newer Prometheus campus in New Albany, Ohio, ABB four-wheel robots equipped with six-axis arms are being tested for tasks like reseating components.

A Kinova Gen3 robotic arm is also being evaluated for controlling power to servers. In another decidedly low-tech application, Meta has used a robotic device that can physically press the power button on equipment such as a Mac Mini after receiving a remote command.

Meta has determined that robots are not currently suited for some demanding jobs, including the extensive cabling required for Nvidia GB300 systems.

Meta has invested heavily in the computing infrastructure needed for AI. The company expects capital expenditures of $130 billion to $145 billion in 2026. In the second quarter, capital spending totaled $31.1 billion, while revenue grew 28% from a year earlier to $60.8 billion.

Automating data center maintenance offers Meta a potential way to control the costs of this infrastructure expansion.

Potential Effect on Jobs

The potential impact on data center jobs is already drawing attention. One Meta worker reportedly estimated that a successful cable-handling robot could take over as much as 80% of the workload for some technicians, though that figure was not an official Meta projection.

Meta maintains that its infrastructure expansion requires more skilled employees. The company has launched a program to train thousands of people each year in electrical and mechanical work, and has partnered with trade unions on apprenticeship programs.

Bottom line: Meta’s work with robots shows how AI-based automation could potentially affect the job market. U.S. tech companies have eliminated nearly 140,000 jobs in 2026, with Oracle, Amazon, Microsoft and Meta accounting for almost 50,000 of those cuts. Certainly other factors played a role in these job losses, including routine cost cutting, but it is likely that AI automation contributed to workforce reduction.