Meta is using tents to house temporary data center capacity, according to a report from SemiAnalysis.

SemiAnalysis reports that Meta is building "multi-billion-dollar GPU clusters" in tents in a bid to get capacity up quicker.

Meta is developing data centers using prefabricated power and cooling modules with "ultra-light structures," and is not installing backup power generation, all in an effort to prioritize speed of deployment.

SemiAnalysis states that Meta is using on-site substations, and "likely uses sophisticated workload management" to maximize efficiency.

A Meta spokesperson has since confirmed the use of tents to house compute capacity.

Tents are not typically an ideal structure for data centers, as the hardware deployed is extremely expensive and generates a lot of heat. SemiAnalysis noted that Meta may have to shut down workloads during the hotter days in summer.

The tent data centers seem to be a temporary solution, however. Dylan Patel, CEO of SemiAnalysis, told Business Insider that they expect Meta to build "full data centers" in the long term, but in the short term, the tents are enabling the company to get capacity up and running "as soon as possible."

Update -

Meta CEO Mark Zuckerberg has since spoken to The Information about the construction method.

“We have a very strong infrastructure team that is doing novel work to build up data centers,” Zuckerberg said. “I wanted them to not just take four years to build these concrete buildings, so we pioneered this new method where we’re basically building these weatherproof tents and building up the networks and the GPU clusters inside them in order to build them faster.”

Fast Company further reports that the tents are being used at Meta's New Albany, Ohio, campus currently under construction.

The tents are made of puncture-resistant and waterproof fabric supported by an aluminum substructure.

Original story resumes -

Meta is investing heavily in what it describes as a "superintelligence" effort. The company is hiring - with CEO Mark Zuckerberg stating, "I'm focused on building the most elite and talent-dense team in the industry" in a post on his Threads platform.

"We're also going to invest hundreds of billions of dollars into compute to build superintelligence. We have the capital from our business to do this," Zuckerberg added.

Also part of that is the company's plans to build several multi-GW clusters - the first of which will be called Prometheus and will come online in 2026.

At the start of this year, Meta said that it would spend $60-65bn on AI data center capex this year, up from $35-40bn the year before.