Meta is testing robots inside its own data centers, the giant server farms that run its apps and its AI systems, to handle physical jobs like unplugging cables, restarting servers, and swapping out parts. The robots come from outside vendors including Kinova, ABB, and a startup called Watney Robotics. One Meta worker who has seen the pilots says a successful rollout could eventually take over up to 80 percent of some employees' daily tasks.
This is not a one-off experiment. Google, Microsoft, and Amazon have all quietly built teams and run trials on data center robotics over the past couple of years, and Microsoft set up a group dedicated to the work back in 2023. Google has specifically looked at robots for moving the heavier server racks used for AI chips, which are far bulkier than older cloud equipment.
The timing makes sense once you look at the money. Meta expects to spend up to 72 billion dollars this year alone building AI infrastructure, and it has told investors that 2026 spending could reach as high as 145 billion dollars. When a company lays out that kind of cash every year, shaving even a small share off staffing costs adds up fast.
Robots that work around the clock without breaks or shift changes look attractive on a spreadsheet, even while they remain slow in practice. A new group of startups has formed to sell exactly this. Watney Robotics raised 21 million dollars last year to build robotic arms for cabling and maintenance work, and a Zurich startup called Exclaim Robotics, founded by a former Google and Microsoft robotics researcher, recently raised close to 5 million dollars for similar machines.
There is a real contradiction sitting underneath all this. Right now, the data center industry says it cannot find enough workers. Two out of three data center operators report trouble hiring and keeping qualified staff, and the industry expects to need well over 100,000 more skilled tradespeople by the end of the decade.
Meta points to this exact shortage as the reason it just launched free training programs for electrical, mechanical, and plumbing work. Both things can be true at once: a shortage of skilled trade workers today, and a steady push toward automation that shrinks how many people each data center needs five years from now.
This matters beyond the workers directly involved. Towns and states have handed out tax breaks worth billions of dollars on the promise that data centers bring lasting local jobs. Research covering hundreds of these facilities found that a typical large data center directly employs only a few dozen to a couple hundred permanent workers.
Even generous studies show a host county gaining only a few thousand jobs after six years. If robots trim staffing further, the jobs argument that officials use to justify tax breaks gets weaker right as public pushback against data centers grows over electricity costs and land use.
If your business depends on data center capacity, through cloud bills, a colocation lease, or a facility of your own, watch for two things moving together: fewer job promises from developers, and quieter arguments from towns about why generous tax deals are still worth it. If you run any large, repetitive physical operation yourself, this is worth studying closely. It shows how automation actually arrives, not through one dramatic announcement, but through years of quiet pilots that only look obvious once you look back.