In one of the more unconventional chapters of the AI infrastructure arms race, Meta has begun erecting large weatherproof tent structures at its data center campus near New Albany, Ohio. The structures, which the company officially refers to as "rapid deployment structures," reportedly allow Meta to house massive arrays of AI accelerator chips while cutting traditional construction timelines in half.
The strategy came to light through findings published by Michael Thomas, founder of infrastructure tracking firm Cleanview. Drawing on local city permit records and satellite imagery, Thomas documented that Meta started building at least five structures measuring 125,000 square feet each between April and June 2026. Each structure has since been completed, with the chips presumably operational inside them.
Meta CEO Mark Zuckerberg had previously hinted at the approach in a conversation with The Information last year, describing a plan to use weatherproofed tent-like enclosures as a faster alternative to conventional data center construction. The announcement at the time raised eyebrows, but the satellite images now confirm that concept has moved well beyond the planning stage.
The tactic draws obvious comparisons to Tesla, which famously assembled Model 3 vehicles inside a large tent in the parking lot of its Fremont factory during a production crunch in 2018. Meta's version is considerably larger in scale: the Ohio site is reportedly powered by roughly 200 megawatts of modular gas turbines, a setup popularized in the AI compute space by xAI, Elon Musk's artificial intelligence company, which has used similar off-grid power arrangements at its Memphis facility.
Why It Matters
The tent-data-center approach is a direct response to a fundamental bottleneck in today's AI buildout: the gap between the industry's demand for compute and the multi-year timeline it normally takes to construct a traditional hyperscale facility. By using pre-engineered structures that can be assembled in weeks rather than years, Meta is effectively trading long-term operational efficiency for short-term speed to deployment.
The approach is not without trade-offs. Tented structures have less robust climate control, are more susceptible to environmental conditions, and require careful cooling engineering to prevent thermal damage to hardware worth billions of dollars. That Meta is willing to absorb those risks reflects the intensity of competitive pressure in the AI model race, where every month of compute availability counts.
For enterprise technology buyers and cloud infrastructure watchers, this development signals that the major AI labs are entering a phase where raw compute accumulation is prioritized above nearly everything else. The pressure on traditional data center real estate, power grid capacity, and cooling equipment is likely to intensify further as competitors respond in kind.