Douglas’s sharpest critique wasn’t about Meta’s strategy—it was about the economics of building AI infrastructure in the U.S. at all, a dynamic he thinks investors are underpricing.

“I think data center capacity in the United States is not going to age well,” he said. “It’s one of the most expensive places to build out that kind of capacity, and a lot of communities don’t want it. I think literally two years from now, those data centers are not going to be very attractive.”

The competition, in his view, is coming from an unexpected direction: sovereign wealth-backed capacity in the Gulf. “The Kingdom of Saudi Arabia is coming online with massive data center capacity, hosted out of Saudi Arabia, at significantly lower prices—because why pump oil out of the ground and ship it overseas in tankers when you can use it here to power massive data centers?” he said, citing Saudi Arabia’s Public Investment Fund, one of the world’s largest sovereign wealth funds. “If I were an investor, no way would I invest in companies building out hyperscaling in the United States right now.”

Douglas added that some Gulf data centers are being structured as legal extraterritorial zones—effectively data “embassies”—so multinational clients with strict data-residency rules can use them without technically moving data outside their home country. “People are focused on putting data centers in a cornfield in Indiana, where no one wants it and it’s very expensive,” he said. “Meanwhile, that story hasn’t really been told.”

Does selling cloud even fit Meta’s business?

Set against that backdrop, Douglas said he was skeptical of the specific plan Bloomberg described—selling excess AI compute the way AWS, Azure, and Google Cloud do. “It doesn’t make a lot of sense unless you really want to put your name in the back of the AI space—just basically get attention,” he said. “Which the announcement is definitely doing. So part of me wonders: is that today’s headline, but tomorrow’s reality?”

Douglas, who spent years as a coding engineer before founding a series of startups, said the plan echoes what SpaceX and xAI have done with their own data centers—building capacity and renting it out to a small number of massive buyers, like Anthropic. That’s a fundamentally different business than the one Meta is reportedly eyeing.

“There’s not many companies that can just take raw data center space and put it to use,” Douglas said. “So there’s a very limited number of customers for that. And if you’re going to rent it like AWS—well, now that’s a very competitive market: AWS, Google Cloud, Azure.”

He was also skeptical that the pivot fits Meta’s core business at all. “Going from three or four billion social media app users to 10 customers buying data center capacity from them—that just doesn’t seem like a good fit,” Douglas said. “But again, we’re sitting here talking about it, which maybe is the point.”

A pattern of swings

Douglas argued the market’s instinct to reward a well-known company for entering a new business—Meta’s stock rose on the report alone—undersells how difficult that pivot actually is.

“There’s often this assumption that if a company that’s successful in one area announces they’re going to enter another area, that’s an automatic success,” he said. “It doesn’t matter what size you’re at—it’s really, really hard to build a new company from scratch, or a new product line, or new revenue, even when there’s proven product-market fit. Meta’s been prolific in going after areas like this—Threads being one example—and they just keep running into it.”

Part of the problem, he said, is talent—”the talent that knows how to build something from scratch usually doesn’t work for a big company like Meta. They’re usually building their own startups.”

Meta’s ads business

Despite his skepticism about the cloud pivot specifically, Douglas said he remains bullish on Meta’s broader AI strategy—particularly its open-source models—and what it means for advertising, even as investors have grown anxious about whether the spending is paying off.

“I’m actually pretty bullish long-term on Meta in terms of AI models and Llama overall,” he said. “I think Meta has the capital, the desire, and the ability to recruit to ultimately do quite well—especially since they’re building and training those models on massive amounts of data and are motivated to make them available at attractive prices. That’s going to benefit their ads business, and it’s going to benefit the ads business generally.” Douglas said MNTN is already in conversations with AI companies about partnering on advertising-specific models for targeting.

He also pushed back on the idea that better ad-targeting AI is bad for consumers. “Consumers pay for everything—companies pay for nothing unless they’re not profitable,” he said. “The better these models can predict who wants to buy what, the lower the cost of the products are. If you make marketing more efficient, that goes back to the consumer in lower prices.”

As for whether a compute-selling business makes Meta look more like Amazon or Microsoft—diversified infrastructure players less reliant on advertising—Douglas said that framing gets it backwards.

“People believe this year Meta will become the world’s largest advertising company. They’re already the world’s largest social media company,” he said, noting the technical feat of inferring user intent from behavior rather than search queries, the way Google does. “Taking a detour toward AI infrastructure—I think it is a detour. It’s not a major strategic move. It’s a tactical move. And all tactical moves have a beginning and an end. I think this one probably would too, if it comes to fruition.”