Meta explores unconventional data center construction for AI growth

Meta’s mad max fix

Meta is racing to add AI computing power by using giant weatherproof rapid deployment structures at its New Albany, Ohio data center campus. Reports citing Cleanview say Meta has built six tent-like structures there, with permits showing five roughly 125,000-square-foot tents that went up between April and June 2026.

The first five traditional buildings at Meta’s New Albany campus reportedly took about two to three years each to complete. The tent strategy is meant to bring computing capacity online faster, and reporters have compared the improvised look to something out of Mad Max.

Inside Meta’s tent data centers

What is expected inside these tents? AI chips are likely worth billions of dollars in total. Public reporting says Meta plans to put large amounts of expensive AI hardware inside the rapid deployment structures, while analyst estimates have placed Nvidia’s GB200 Superchip at roughly $60,000 to $70,000.

That math explains why the setup has drawn so much attention. If tens of thousands of high-end chips are involved, the value under these fabric-supported structures could reach several billion dollars, even before counting networking, cooling, power, and other infrastructure.

Power plants in the backyard

Meta’s New Albany data center expansion is tied to dedicated behind-the-meter power. In June 2025, the Ohio Power Siting Board approved the 200-megawatt Socrates South Power Generation Project, a natural gas-fired facility designed to serve a Meta-affiliated data center customer.

The project is being built and operated by Will-Power OH, a subsidiary of Williams Companies. Filings described the plant’s output as dedicated to a single customer’s electric load and not physically connected to the electric grid, with fuel supplied by two 24-inch natural gas pipelines.

Borrowed from Elon Musk

Meta’s rapid deployment strategy is being compared with two Elon Musk playbooks. Tesla used a tent-like Sprung structure at its Fremont factory in 2018 to add Model 3 assembly capacity quickly, and Sprung lists that structure as 915 feet long, 150 feet wide, and 53 feet tall.

xAI’s Colossus project in Memphis has also become a reference point for speed. Business Insider reported that the supporting facility and supercomputer were built in 122 days, while Musk said it took 19 days to move from hardware installation to training. NVIDIA CEO Jensen Huang said a project like Colossus would normally take years.

Fun fact: Tesla’s Fremont tent was 53 feet high and 150 feet long with aluminum framing. It helped Tesla reach a production goal of 5,000 Model 3 vehicles per week by late June 2018.

Why tents instead of buildings?

Building traditional data center capacity can take years, especially when permits, utility connections, concrete, steel, and large construction teams are involved. In New Albany, Meta started building five roughly 125,000-square-foot rapid deployment structures between April and June 2026, and satellite images reviewed by Cleanview showed the structures had already been built.

Speed is the reason this approach matters. The first five traditional buildings at Meta’s New Albany campus reportedly took about two to three years each, while the rapid deployment structures were put up far faster.

Fun fact: About 30% of all planned data center capacity in the United States is now designed to use behind-the-meter power, bypassing the public grid entirely.

Billions under canvas

The rapid deployment structures are expected to hold AI chips likely worth billions of dollars in total. That makes the New Albany setup unusual: expensive computing hardware is being placed inside fabric-supported structures instead of conventional concrete-and-steel data center buildings.

The computing power is meant to support Meta’s broader AI push. Public reporting has also tied Meta’s infrastructure race to Muse Spark, its newer AI model whose developer API had reportedly faced delays after the model was announced in April 2026.

The Tennessee expansion

Ohio is not the only place where Meta is using this rapid deployment approach. Reporting citing Cleanview says Meta has already built or is constructing three data centers that use the strategy, including one site in Tennessee.

The Tennessee expansion suggests Meta is testing rapid deployment structures beyond New Albany. Public reporting supports the broader pattern of tent-like structures and dedicated power strategies, but the full site-level details for every location have not been publicly documented in the same way as the Ohio project.

Avoiding community backlash

Communities across the United States are raising concerns about new data centers, including electricity demand, water use, land use, noise, and limited long-term job creation. Behind-the-meter power can help companies reduce dependence on utility interconnection timelines, but it can also create new scrutiny over gas generation, emissions, and local impacts.

In New Albany, Ohio, regulators approved the 200-megawatt Socrates South project to directly supply a Meta-affiliated data center. The approval shows how AI infrastructure is increasingly being paired with dedicated power, but not that local concerns disappear.

$145 billion bet

Meta expects 2026 capital expenditures, including principal payments on finance leases, to reach $125 billion to $145 billion. That is far above the $72.22 billion it reported for full-year 2025 capital expenditures.

Wall Street has been watching the spending closely, and TechCrunch reported that Meta’s stock was trading down about 5% for the year as the capital bill climbed. Rapid deployment structures are one way Meta may reduce construction time and cost while trying to add the compute capacity it needs to compete with OpenAI, Google, and other AI rivals.

The Mad Max vibe

One reporter looked at these gas-powered tents and said it’s like a scene out of the movie Mad Max. You know, the post-apocalyptic film where everything is gritty and improvised. That’s exactly what this is: an improvised AI infrastructure.

When you can’t build fast enough the normal way, you strap jet engines to the ground and throw up some tents. Michael Thomas, founder of Cleanview, called this the Mad Max phase of the AI race. It’s chaotic, it’s weird, but it’s also kind of impressive. Desperate times call for desperate measures.

What about security?

The security question is part of why Meta’s tent strategy has drawn attention. Tom’s Hardware cited a commenter who compared the setup to a “$10k racing bike with a $9 lock,” a vivid way to describe concerns about placing expensive AI hardware inside fabric-supported structures.

Public reporting does not detail Meta’s full security setup for the New Albany structures. What is clear is that Meta is using rapid deployment structures to bring AI compute online faster than conventional data center construction would allow.

A growing trend

Michael Thomas and Cleanview report that about 2 gigawatts of behind-the-meter generation capacity is online as of mid-2026. Cleanview says that if projects with signed tenants meet their timelines, cumulative operating capacity could reach 13 gigawatts by the end of 2027, though the report also warns that delays could leave the total much lower.

That 13-gigawatt scenario would exceed New York City’s power demand. Cleanview’s June 2026 report also found that 92% of the behind-the-meter projects it identified were announced since the beginning of 2025, showing how quickly dedicated power has moved from a niche strategy to a major data center trend.

If you’re curious about how the rapid expansion of tech infrastructure is creating new challenges beyond just power and data centers, check out how court rules require Meta to face a Massachusetts youth addiction lawsuit for another example of the growing scrutiny facing the industry.

The bottom line

Meta’s tent data centers show how far the AI infrastructure race has pushed major tech companies. High-end AI chips are expensive, power-hungry, and in huge demand, while traditional data center construction and utility connections can move much more slowly than AI companies want.

The rapid deployment structures are a shortcut, not a normal-looking data center strategy. They give Meta a faster way to add compute capacity at a time when AI competition with OpenAI, Google, xAI, and others is intensifying.

If you want to see how Meta is reshaping more than just its infrastructure to stay competitive in the AI race, check out how Meta job cuts spread across multiple divisions during restructuring for a closer look at the company’s broader transformation.

If you found this slideshow interesting, hit that like button and drop a comment below. We’d love to hear your thoughts on Meta’s tent-powered AI gamble.

This slideshow was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content right here on MSN.

Read More From This Brand:

• Microsoft faces backlash for 6000 visa push amid job cuts
• Verizon plans to cut over 13,000 jobs in its latest restructuring move
• Amazon announced 14,000 job cuts in an effort to streamline global operations