In early 2023, Blackstone CEO Steve Schwarzman gathered with Jas Khaira, the company’s lead AI investor, and other top executives at the finance giant’s Park Avenue headquarters in Manhattan for a debate: Should they give billions of dollars to a little-known data center developer—CoreWeave—that they had turned down just a few months earlier?

CoreWeave had gotten Blackstone to return to the table after landing a multiyear, multibillion-dollar contract to sell cloud computing services to Microsoft. It needed $2.3 billion to buy Nvidia GPUs and secure a data center to house them.

In Blackstone’s discussion of the deal, several executives in the room had deep misgivings, according to a person who attended the meeting. CoreWeave was still something of a mystery, a six-year-old, privately held firm that had generated only $16 million in revenue in the previous year. Schwarzman himself was personally worried his investment team could be making a mistake, the person said.

Secretive Wall Street trading firm Jane Street has become famous in the past couple of years for its sizable AI-related spending, as it has struck big cloud deals to handle its AI computing needs and made investments in neocloud CoreWeave. Yet even a Jane Street–related deal for data center debt hasn’t won over investors.

Debt sold by a venture building a data center to be leased by Jane Street has quickly soured in secondary trading, according to Financial Industry Regulatory Authority data. Yields on the bonds have risen to around 11.3%, more than 2 percentage points above where the debt first sold in August. That deterioration has been sharper than for other recently issued AI data center bonds, with investors demanding far higher compensation to hold them versus government debt.

Greg Jensen, the managing chief investment officer of Bridgewater Associates, one of the world’s largest hedge funds, said demand growth for AI infrastructure is largely priced into markets. As a result, Bridgewater has begun looking for investment opportunities elsewhere.

“We thought this was ...

Greg Jensen, co–chief investment officer at Bridgewater Associates, thinks the top holders of AI compute warrant extra regulation, similar to what governs the world’s biggest financial institutions.

If a company controls more than, say, 5% of U.S. or global compute, it should be subject to heightened oversight, “like we do with systemically important banks,” Jensen said. In a wide-ranging interview, he also discussed Bridgewater’s growing use of AI in investing and where it still remains cautious, including leaving risk controls and trade execution to what he called “human-driven algorithms.”

In May, Blackstone unveiled a new joint venture with Google that revolved around a big number: The Wall Street giant was promising to spend $5 billion to buy 500 megawatts’ worth of Google’s tensor processing units, and it said it would use the chips in a new company that would deliver computing power to leading AI firms.

Four months later, Blackstone’s plans have only gotten larger, according to a person with knowledge of the effort. Blackstone actually anticipates buying many more TPUs—several “multiples” beyond the originally announced figure, the person said.

In other words, Blackstone is likely to commit tens of billions of dollars to securing multiple gigawatts of AI chips. Blackstone declined to comment on exactly how many more TPUs it’ll buy.

Jas Khaira, the leader of a Blackstone team dubbed N1 that is leading the company’s push into AI investments, said in an interview that the firm’s “biggest conviction bet continues to be compute.” The company, he said, views the infrastructure build-out as something that will be bigger than a simple technology cycle. “This is definitely more akin to an industrial revolution,” he said.

Anton Levy, the former co-president and chair of the global technology group at General Atlantic, has raised more than $1.1 billion for his inaugural fund, a sign that he has been able to compete with some of the biggest names in venture and growth equity during a challenging environment for ...