A potential pullback by Middle East sovereign wealth funds could drain hundreds of billions of dollars from the artificial intelligence boom and threaten key data center projects, according to tech investor Jack Selby.

Middle East investors — including sovereign wealth funds and government entities — account for roughly a quarter of global investments committed to AI over the next five years, said Selby, managing director of Peter Thiel’s family office, Thiel Capital. If the war in Iran drags on, and the United Arab Emirates, Saudi Arabia and other countries divert their investments to rebuilding at home, the lost capital could ripple through data centers as well as public and private tech companies, he said.

“I think markets have underappreciated how important the Middle East region is for capex spending as it relates to AI and AI infrastructure,” Selby told CNBC in an interview. “If the Middle East starts taking some of these projects offline or canceling some of these projects, the impact on the market could be much, much, much larger than what they currently suggest.”

Selby’s warning has implications for high-net-worth investors, family offices and funds betting on the AI trade. A Wall Street Journal report this week about missed revenue targets at OpenAI rattled tech and chip stocks. Selby said the Middle East poses another funding risk, as AI companies grew more dependent on the region for capital.

Oracle, Nvidia and Cisco are part of OpenAI’s campus in the UAE to build out 5 gigawatts of capacity. Microsoft plans to invest $15 billion in the UAE by 2029. The sovereign wealth funds of the UAE and Saudi Arabia have become key investors in private AI companies, with OpenAI reportedly seeking $50 billion from the big funds in the region earlier this year.

Selby estimates that half of the Middle East’s AI funding is dedicated to data centers located in the region. The other half is allotted to projects and data centers worldwide. Middle East funds and companies have already started canceling various shipping and business contracts by invoking force majeure, he said. The big risk is that they start canceling data centers as well.

“Markets don’t seem to grasp that this is a very real situation,” he said. “It’s very volatile. I hope and I pray that it goes back to some semblance of normalcy soon. But it seems to me that markets are underpricing this volatility and the risk.”

Beyond the war, AI also faces a broader risk of overinvestment and speculation, Selby said. Like the dot-com bubble, he said investors and founders are bidding up values of AI and infrastructure companies indiscriminately. He said the AI boom is consuming far more capital, with the top hyperscalers expected to spend more than $700 billion this year. So the wealth destruction will overshadow the losses of the dot-com bust.

“AI is a revolutionary technology, don’t get me wrong,” he said. “But it can also be an exceptional bubble. There will be extreme winners and there also be some real losers. And those losers will be orders of magnitude larger than any of the losers that we’ve seen before. The AI bubble, when it busts, will be at least one more zero, probably two and three more zeros than the dot-com bubble. That will be tens, if not hundreds, of billions of dollars.”

He cited Google as an example from the dot-com era. While investors were bidding up the values of Ask Jeeves, Infoseek, AltaVista and other early search functions, Google came along and upended all their business models. He said similar disruptions could happen to today’s AI leaders.

Selby’s AI strategy is to avoid the crowds. With a second fund he’s launching at Copper Sky, his Arizona-based VC fund, Selby is targeting tech firms outside of California, New York and Massachusetts. He said tech firms in those three states — especially the Stanford and MIT clusters — are attracting all the capital and attention. So the best values lie elsewhere, he said.

“Probably 90%-plus of all venture capital investment went to California, New York, Massachusetts, an all-time high,” he said. “The good news is you get outside of those three states and go to the other 47 states, the deals, the investment opportunities are far, far, far less expensive, and that’s what we do.”

Selby declined to give many details on Thiel’s family office, saying only that Thiel invests in great founders rather than specific industries. Thiel Capital, which ranked on the Inside Wealth Family Office 15 list of most active family office investors, has invested in everything from German drone makers (Stark) and  gene therapy startups (Kriya Therapeutics) to an AI hiring company (Mercor) and space research firm (Varda).

Yet as a family office director and head of a VC fund that raises money from family offices, Selby said the biggest mistake for many family offices today is making their own direct investments. A survey from Citibank last year found that seven out of 10 family offices have made direct investments in private companies, without going through a fund.

Selby said he understands why family offices are striking out on their own, given the dismal performance of private equity and venture capital funds and lack of distributions. He said two-thirds of venture capital firms are “zombie VCs,” that aren’t raising or returning money and should close.

“Family offices are so frustrated with people like ourselves, who have not been returning their capital, so why shouldn’t they try it themselves?” Selby said. “They couldn’t do any worse than a lot of what [VCs] have been doing in terms of making investments, not giving money back, having marks on paper.”

At the same time, however, he said typical family offices aren’t adequately trained in assessing, valuing and restructuring private companies. Many ultra-wealthy investors are more motivated by status and peer pressure than by disciplined returns.

“When these fancy people go to their cocktail parties in Manhattan, they have to have something interesting to talk about,” he said. “All of their friends are talking about some version of [direct investments]. So they have to have something to add to the conversation. So therefore, they do the same thing. The Greek shipping magnate that lives in Manhattan knows nothing about rocketry. So why is he investing in SpaceX? Because he just wants to have something fun to talk about at the fancy cocktail party.”

The ⁠Food and Drug ⁠Administration on Thursday proposed ​excluding the active ingredients in Novo Nordisk ‌and Eli Lilly‘s blockbuster obesity and diabetes medications from the ⁠list ⁠of drugs that outsourcing facilities can use for ​compounding in bulk.

If that proposal is finalized, the exclusion would likely limit the mass compounding — or the making of custom, often cheaper alternatives — of those medicines ​unless they appear on the FDA’s ⁠drug shortage ⁠list. The agency said it will consider public comments, which can be submitted until late June, before making a final decision.

The FDA finds “no clinical need” for outsourcing facilities to compound them from bulk drug substances, the agency said in a release.

The proposal includes semaglutide, the active ingredient in Novo’s obesity drug Wegovy and diabetes counterpart Ozempic, and tirzepatide, which is in Lilly’s weight loss injection Zepbound and diabetes shot Mounjaro. It also covers ⁠Novo’s older molecule liraglutide.

“When FDA-approved drugs are available, ​outsourcing facilities cannot lawfully ​compound ⁠using bulk drug substances unless there is a clear clinical ⁠need,” ​FDA Commissioner Marty ​Makary said in the release.

The agency’s proposal specifically targets 503B outsourcing facilities, which manufacture compounded drugs in bulk with or without prescription and are largely regulated by FDA guidelines. 

The proposal does not impact 503A pharmacies, which make compounded drugs according to individual prescriptions for a specific patient and are largely regulated by states rather than the FDA. 

Lilly and Novo have invested billions to ramp up manufacturing capacity over the last several years, which has helped alleviate supply constraints. The companies have also pursued efforts to make their branded medications more affordable to win over users who had flocked to cheaper compounded medications.