Robots Could Soon Take Over 80% of Meta’s Data-Center Workload – and There’s a $145 Billion Reason Why

Listen to the audio version of this article (generated by AI).

Meta Platforms (META) is doubling down on the use of robots to handle a variety of tasks in its data centers… tasks once carried out by humans. If you think this sounds like the premise of a dystopian sci-fi movie, you’re not alone. Just look at what one Meta data center worker told a reporter from Wired:

“We thought those of us performing the physical tasks were safe for a while, but not anymore. It’s coming for us all, unfortunately.”

Artificial intelligence (“AI”) has already claimed hundreds of thousands of white-collar jobs.

And, as Meta’s robots have recently demonstrated, it’s coming after blue-collar workers as well. The same worker behind that quote you just read believes that the Watney Robotics cable-swapping robot Meta is testing could perform up to 80% of some employees’ workloads.

While that thought is disturbing enough, the reason behind Meta’s quest for more automation in its data centers is even harder to swallow.

What’s Happening Inside Meta’s Data Centers?

A 2023 YouTube video demonstrated Meta’s push for automation. In the video, self-driving tugger robots move massive server racks inside its data centers, and wheeled robots read barcodes to log inventory.

But that was just the beginning. Meta has since begun using finger-like robots to restart devices when a human prompts them remotely. The company has also tested robots that can reset servers, plug in cables, cycle power, cut off electricity to servers, and, as noted earlier, swap out old networking cables for new ones.

At its new Prometheus data-center complex in New Albany, Ohio, Meta has been testing four-wheel robots with a riser (like a scissor lift) topped by a six-axis arm. This robot, built by ABB (ABBNY), is being used to reseat loose hardware components.

Are these high-skill, specialized tasks? Not especially. But humans have always performed them as part of their day-to-day responsibilities. With robotics technology seemingly advancing by the day, the question isn’t whether robots will be able to perform those high-skill, specialized tasks… It’s when.

Meta is trying to answer that question by testing various robots built by manufacturers like ABB, Watney, and Kinova.

And while they’re not yet ready to operate completely autonomously, data-center workers believe they’re getting closer. That has many Meta employees concerned about their livelihoods.

And that wasn’t supposed to be part of the plan.

The Irony Behind Meta Turning to Robots as Data-Center Workers

The irony surely isn’t lost on the Ohioans, Iowans, Oregonians, and Louisianans who reside in or near the towns of New Albany, Altoona, Prineville, and Richland Parish, respectively, where Meta’s largest U.S. data centers are located (or will be located, with Louisiana’s Hyperion campus scheduled for first-phase completion by 2030). It’s especially not lost on those who work in these facilities.

It’s no secret that a fast-growing wave of communities across the country do not want data centers in their backyards. Public backlash continues to grow, and one of the carrots that hyperscalers and data-center operators dangle in front of these communities is the promise of thousands of local jobs within or related to the data centers.

Job creation is especially appealing for the construction workers and contractors tasked with building these AI campuses. That’s real, tangible work – but it’s temporary. Once construction is complete and data centers are operational, Meta will need human employees to work within the buildings and campuses.

While pitching its data centers to these communities, Meta promised thousands of jobs would be created:

• Louisiana: 1,000 permanent positions at the Hyperion campus in Richland Parish.

• Ohio: Roughly 400 positions between the Prometheus campus in New Albany and a new facility in Bowling Green.^,

• Iowa: More than 400 jobs at Meta’s Altoona campus.

• Oregon: More than 350 positions at the Prineville facility.

It’s hard to say whether the number of actual employees that are (or will be) working in these data centers matches Meta’s promised numbers. But it’s likely in that ballpark, or a bit less.

Aside from the promise of jobs, however, hyperscalers like Meta offer plenty of other sweeteners to win over small-town and rural communities like these so they can build their data centers.

For example, Meta has provided more than $19 million in direct funding to schools and nonprofits in the counties and/or surrounding areas that house the Altoona, Prineville, and New Albany data centers. The company has also provided more than 800 local grants and sponsorships in these areas.^{, ,}

Meta is certainly investing in these communities, and they are noble endeavors. But the permanent jobs are the primary appeal and selling point for folks living in these areas.

If those jobs are in danger, the data centers lose their luster among these communities. And, more importantly, in areas where they want to build new data centers… because no one wants to be a bait-and-switch victim – especially of Meta’s push toward more automation.

$145 Billion in Spending Pushes Meta to Cut Costs and Automate Operations

Global AI investment, much of it from hyperscalers like Meta, Amazon (AMZN), Microsoft (MSFT), and Alphabet (GOOGL), is projected to reach more than $1 trillion in 2026, according to Goldman Sachs. Meta had originally estimated a 2026 floor of $115 billion in AI spending before raising it to a range of $130 billion to $145 billion. It’s simply mind-blowing.

Unfortunately, when a company – even one with as much capital as Meta – decides to spend at this level, there are inevitably cutbacks. Especially when its year-over-year net income drops by roughly 13.6%, its operating margin falls by 12%, and its diluted earnings per share drop 13%, as they all did during the second quarter of 2026.

It’s no secret that Meta has already laid off a significant chunk of its workforce this year – 8,000 jobs were cut in May, and another 6,000 open positions were canceled in April, amounting to a 14,000 reduction in headcount.

After all, data centers cost a lot of money to operate. One late-2025 estimate determined that a typical 100-megawatt hyperscale data center or AI campus costs upwards of $1 billion to run… per year.

Here’s the thing, though – since Meta’s data centers only employ a few hundred people, how much money is the company actually saving by automating some jobs? Maybe $15 million to $20 million? It’s a drop in the ocean for Meta.

The key driver is downtime. That’s what keeps data-center operators up at night. Research has shown that one minute of unplanned IT downtime can cost large enterprises up to $23,750. The same research calculated that average downtime lasts between 30 minutes and two hours. That’s a potential loss of $712,500 to $2.85 million.

There’s a lot of money at stake. At least 50% of organizations reported one or more significant data-center outages over the previous three years in 2025 and that 54% of the respondents said their most recent significant, serious, or severe outage cost more than $100,000 (41% reported hourly downtime costs above $1 million).

Beyond the cost per minute of downtime, other financial factors come into play, including:

• Potentially debilitating Service Level Agreement penalties – often ranging from 15% to 500% of the data operators’ monthly base rent, depending on the severity of the outage – if promised uptime levels aren’t met.

• Lost employee productivity if workers have to temporarily abandon their duties to fix the outage.

• Customer dissatisfaction and reputational damage, both of which are hard to move past.

Alarmingly, Uptime Intelligence’s annual outage analysis 2024 report estimated that human error – directly or indirectly – contributes to anywhere between two-thirds and four-fifths of all downtime incidents (not specific to data centers).

Specific to data centers, nearly half of organizations (49%) either confirmed or weren’t sure whether significant, serious, or severe IT outages were caused by human error.

These figures undoubtedly strengthen Meta’s argument for automation.

Robotics Will Help Automate Some Tasks, but There Are Limits

The stark reality is that some human responsibilities within data centers can be handled as well – if not better – by robots. For example, instead of a human technician tracking down an issue across millions of square feet of equipment, a robot can quickly diagnose and swap out the broken component.

But automation has its limitations. Meta is using the inventory robot it’s testing to inspect equipment failures. But it can’t yet distinguish between green and red indicator lights on the equipment since it only “sees” in grayscale. That requires human intervention.

Meta has also acknowledged that its robots are not yet ready to handle certain tasks, such as the complex cable configuration and hardware of the Nvidia (NVDA) GB300 supercomputers used in some of Meta’s data centers.

At a basic level, that same robot runs into mobility issues because its wheels can’t navigate around cables on the floor. The robot also struggles to round corners. There’s also the downtime needed for the robots to recharge, plus the human inconvenience of moving robots from one building to another. This is clearly a work in progress.

But it might be a necessity. Not only so Meta can reinvest that saved money into AI, but also because there simply aren’t enough skilled trade workers to fill positions at its data centers. Meta has stated publicly that it’s looking for more of those workers, not fewer.

That doesn’t quite mesh with what Meta’s senior manager for robotics, Eric Xu, envisions. He says Meta’s long-term goal is to install robots in data centers because they could address incidents faster, automatically monitor the data centers, and handle routine maintenance.

Robotic automation is a sensitive topic that will always have its proponents and detractors. The pro-automation faction argues that robots are more efficient, more consistent, and remove the possibility of human error. These folks also contend that robots are necessary because there aren’t enough qualified local workers.

Proponents of automated work also argue that robots will actually help workers by taking on tedious and even dangerous tasks, freeing up time for humans to focus on more complex work. This has been a common refrain from the pro-AI crowd, who insist that AI bots help human workers, rather than harm them.

There may be some truth to that, but if a robot will soon be ready to perform most of a human’s job duties, it’s hard to see how that helps Meta’s human workforce.

Meta Robotics Developments to Watch Into 2027

As usual, it all comes down to money. If Meta continues to spend on AI as it has, will it cut headcount further? Let’s not forget, however, that cutting its workforce doesn’t save Meta much money in the big picture.

Here’s what I wrote in my May 29 article about Meta’s massive layoffs:

To be blunt, the layoffs had nothing to do with boosting the near-term bottom line and everything to do with reinvesting those costs into AI. Ironically, many of those laid-off workers contributed to the creation of the technology that rendered their own roles expendable.

The layoffs and the spending are all about building Meta’s AI machine. So, keep an eye on the company’s capital expenditures (“capex”) during its third- and fourth-quarter earnings calls (October 28 and late January 2027, respectively) to see if it surpassed earlier projections.

Also, how will Meta’s stock react to capex spending news? Investors are extremely wary of major AI spending at this point.

Meta’s stock has been slightly down this year. But there’s still some room to grow. As of September 3, Meta’s average price target of $753.60 represents a 23% upside on its trading price of $610.68.

It’s also important to track the progress and deployment of not only the robots currently in Meta’s Altoona and New Albany data centers, but also in its other data centers across the country.

If Meta can solve the current issues plaguing its robots, there’s no reason to think they won’t be deployed on a large scale, even as those robots negate Meta’s pitch and promise of more jobs to potential data-center sites.

Anything to gain an operational edge as Meta continues funneling money into its AI build-out.

Regards,

David Engle

Editor’s Note: Whitney Tilson called the rise of Apple, Amazon, and Netflix… as well as the collapse of dozens of companies that went bankrupt. Now the former $200M hedge fund firm manager is stepping forward with what he calls the most important financial warning of his 30-year career. He’s sharing two free stock recommendations  (one to buy, one to sell immediately)  along with details of a new proprietary system fueling his predictions. See it all in his free presentation.