With questions swirling about whether hundreds of AI startups will ever translate the tech into real-world economic returns, new players in the space are scrambling to justify all the hype. One of the particularly baffling ways they’re doing that is by renting attractive, grown-up office spaces in Manhattan — even if their business really only exists on a couple of MacBooks.

Speaking to the Wall Street Journal about the bizarre AI real estate market, vice chairman of New York brokerage at JLL Benjamin Bass said that AI startups are leasing offices vastly larger than their staff numbers actually require — and that’s without getting into the fact that these are software-driven companies where pretty much all the work can be, and frequently is, done from home.

For example, one AI health startup called Adonis leased a 25,000 square-foot office at 3 World Trade Center, the second tallest skyscraper at the office complex. Assuming the most liberal use of space at 250 square-feet per worker, that kind of office is fit for a headcount of 100, at a minimum. Yet according to the WSJ, Adonis had just 25 employees when it signed the lease. That number has grown, but only to about 50 to 60 in-office workers.

“We were basically giving the company a size-12 shoe, and we were size-4 at the time,” co-founder and CEO Akash Magoon told the paper. “We figured it would be motivating.”

While some AI companies are trying to secure long-term rentals in the ultra-competitive landscape of New York City, others have found four walls a requirement to passing from startup to legitimate business.

“We had a customer literally ask us if we were in-person as part of their diligence process,” Caitilin Leksana, CEO of the AI startup Fazeshift told the WSJ. (Despite having a team of only around a dozen people, Fazeshift recently opened their second office in Manhattan.)

Whatever the reason, it’s clear these companies are all flush with gobs of cash from their venture capital backers, which continues to subsidize the vast majority of AI companies — and while the money spigot’s flowing, they’re busy trying to spend it in whatever flashy ways they can.

More on AI companies: Dark Cloud Gathers as Bill Comes Due for AI Industry