Meta accelerates AI infrastructure with tent-based data centres amid cost pressures
Construction of six weatherproof facilities outside New Albany reduces build time, but stock faces headwinds as Muse Spark release delays persist
Meta has constructed six temporary data centres in weatherproof tents outside New Albany, Ohio, a move designed to slash construction timelines and reduce costs for its expanding artificial intelligence infrastructure. The initiative, which involves housing billions of dollars worth of AI chips in 125,000-square-foot structures, mirrors unconventional deployment tactics previously employed by Tesla and xAI.
The project was identified by Michael Thomas, founder of Cleanview, through a review of local permits and satellite imagery. According to city documents, Meta began construction on five of these 125,000-square-foot structures between April and June 2026. Satellite images shared by Thomas on X confirm that the initial five structures have been completed, with reports indicating the company is building dozens of similar facilities across US campuses.
The strategy appears to borrow heavily from competitors in the sector. Tesla previously utilised tents in the parking lot of its Fremont, California factory to expedite the rollout of the Model 3, while xAI has widely deployed modular gas turbines to power its data centres. The Ohio site is powered by 200 megawatts of modular gas turbines, aligning with the off-grid power solutions favoured by xAI.
This acceleration in physical infrastructure comes as Meta faces internal and external pressure regarding its software development. Reports indicate that while Meta’s latest model, Muse Spark, is complete, the APIs required for developers to access the large language models have been repeatedly delayed. The company had previously discussed plans to use weatherproof tents for multi-gigawatt data centres in an interview with The Information last year.
The construction efforts are set against a backdrop of significant financial scrutiny. Meta intends to spend up to $145 billion on data centres and other capital expenditures this year. The scale of this spending has weighed on investor sentiment, contributing to a 5% drop in Meta’s stock price over the past year. By utilising rapid deployment structures, Meta aims to trim the bill associated with its massive AI build-out.
TechCrunch has reached out to Meta for comment regarding the specific details of the New Albany site and the status of the Muse Spark API release.
