Meta is taking an unusually fast and industrial approach to its artificial intelligence buildout, using large tent-like structures to house AI computing hardware as it races to bring new data center capacity online.
The Facebook and Instagram parent has built a series of massive temporary-style structures near New Albany, Ohio, as part of a broader effort to shorten the time required to deploy AI infrastructure. The structures, described in permitting documents and industry tracking as “rapid deployment structures,” appear to mark a major shift from the concrete-heavy, multi-year data center construction model that has defined the cloud industry for more than a decade.
Meta Turns to Rapid Deployment Structures
The strategy has drawn comparisons with Tesla’s use of large tent structures during the Model 3 production ramp in Fremont, California, when the electric-vehicle maker turned to temporary production space to accelerate output. For Meta, the urgency is not cars, but compute. The company needs vast amounts of graphics processing capacity to train and run advanced AI systems across Facebook, Instagram, WhatsApp, its AI assistant, advertising tools, and future consumer devices.
The latest public data suggests the Ohio project includes six large tent-style buildings, with several measuring about 125,000 square feet each. Data center tracker Michael Thomas, founder of Cleanview, said public permits and satellite imagery show that Meta started building five of the structures between April and June 2026 and that the buildings have already been erected.
The speed is significant because permanent data center buildings can take years to complete, especially when land preparation, power connections, cooling systems, permitting, and supply-chain constraints are included.
Zuckerberg Pushed for a Faster Buildout
Meta Chief Executive Mark Zuckerberg has previously framed the approach as a way to avoid waiting years for traditional data center construction. In comments reported last year, Zuckerberg said, “I wanted them to not just take four years to build these concrete buildings.”
He added: “So we pioneered this new method where we’re basically building up these weatherproof tents and building out the networks and the GPU clusters inside them in order to build them faster. They are hurricane-proof tents.”
That quote captures the central bet behind Meta’s infrastructure push. The company is not merely adding more data center space. It is trying to change the rhythm of construction so expensive AI chips can be put to work sooner. In the current AI race, idle chips are a major cost. A graphics processor that sits in storage while a building is finished is not training models, serving users, or improving products.
AI Spending Is Driving the Infrastructure Race
The Ohio tents are part of a larger wave of AI-focused data center expansion across the United States. Meta has told investors it expects 2026 capital expenditures, including finance lease principal payments, to reach between $125 billion and $145 billion. The company said the increase reflects higher component pricing and additional data center costs needed to support future capacity.
That figure puts Meta among the most aggressive spenders in the AI infrastructure race, alongside Microsoft, Alphabet, Amazon, and xAI. The company is investing heavily to support its AI assistant, recommendation systems, advertising products, generative AI tools, and future hardware ambitions.
The tent strategy also shows how data center economics are changing. Traditional cloud facilities were built for reliability, efficiency, and long operating life. AI facilities still require those qualities, but they also face a new constraint: speed. Frontier AI development depends on how quickly companies can assemble clusters of GPUs, high-bandwidth networking, cooling, and power.
Power Supply Becomes a Key Bottleneck
Power is a central part of the Ohio buildout. Thomas has said Meta signed a long-term deal with Williams to support a pair of 200-megawatt off-grid power plants for the rapid deployment structures. That points to another pressure point in the AI boom.
Big technology companies are no longer only competing for chips and land. They are competing for electricity, grid access, transformers, gas turbines, substations, cooling equipment, and construction labor.
The use of off-grid or behind-the-meter power has become more attractive because utility interconnection queues can take years. For AI companies trying to deploy massive clusters quickly, waiting for the grid can be as limiting as waiting for a building. By pairing temporary-style structures with dedicated power sources, Meta appears to be reducing two of the biggest bottlenecks at once: construction time and power availability.
The Model Comes With Operational Risks
The approach is not without risk. Housing billions of dollars of AI chips inside fabric or tent-style structures raises questions about resilience, cooling, maintenance, insurance, fire safety, and long-term reliability. Zuckerberg has described the structures as weatherproof and hurricane-proof, but the model is still unconventional at this scale.
AI clusters produce large amounts of heat, and even small interruptions in cooling or power can disrupt expensive workloads. That makes the engineering challenge more complicated than simply putting servers under a roof. The buildings must still support security, networking, cooling, backup systems, and reliable operations.
Environmental questions are also likely to follow. AI data centers are already under scrutiny for electricity use, water consumption, emissions, and pressure on local grids. If rapid-deployment facilities rely on gas turbines or other dedicated fossil-fuel generation, critics are likely to ask whether speed is being prioritized over cleaner energy planning.
Meta Tries to Compress the AI Timeline
For investors, the tents tell a second story: Meta is trying to make its AI spending more productive. Wall Street has become increasingly sensitive to the size of AI capital expenditure across the technology sector. Meta’s advertising business remains highly profitable, but the company’s spending plans have raised questions about when AI investment will generate measurable returns.
Faster deployment could help Meta defend the argument that its heavy spending is not simply defensive, but operationally urgent. If the company can bring AI clusters online months earlier, it can use expensive chips sooner and improve the return on infrastructure that would otherwise sit idle.
The competitive backdrop is important. OpenAI, Google, Anthropic, xAI, Microsoft, and Amazon are all racing to secure chips and power. xAI’s fast buildout of its Colossus supercomputer in Memphis showed how quickly an AI company can move when it prioritizes speed and dedicated power. Meta’s tent structures suggest that Zuckerberg does not want the company to lose ground because of slow construction cycles.
Why the Tesla Comparison Matters
The comparison with Tesla is useful because both cases involve companies trying to bypass conventional infrastructure timelines. Tesla used tents when factory capacity became a constraint. Meta is using a similar logic as computing capacity becomes the bottleneck for AI.
The physical appearance may look temporary, but the business logic is serious: build faster, deploy sooner, and keep expensive hardware active. The move also signals that the AI boom is becoming less like a software cycle and more like a heavy industrial race. The winners will not be decided only by algorithms, researchers, or product design. They will also be shaped by real estate, energy contracts, modular construction, supply chains, and the ability to manage complex infrastructure at extreme speed.
What Comes Next
Meta’s tent-based data centers may not become the default model for the industry, but they show how far major technology companies are willing to go to compress timelines. In the AI race, a traditional data center that takes years to complete may be too slow for the current market.
For now, the Ohio tents stand as one of the clearest symbols of the AI infrastructure scramble. Meta is not waiting for the old data center playbook to catch up. It is rewriting parts of that playbook in fabric, steel frames, gas turbines, and billions of dollars of chips.
Comments