Private equity giant Blackstone has positioned itself at the center of the AI boom, company veteran Jas Khaira said on stage at TechCrunch Disrupt last week.
The company is cutting checks through a variety of strategies, from providing huge loans for data centers to investing in cutting-edge AI models and applications.
Earlier this year, the company launched Blackstone N1 (BXN1), a central hub for its AI initiatives based in San Francisco. The group is led by Kaila, who immigrated from New York to become an AI hub.
Unlike previous technology cycles such as mobile and SaaS, where companies were able to reach great scale with modest capital, the transition to AI requires huge amounts of investment. The need for capital is why Blackstone, with $1.3 trillion in assets under management, can play a key role in this technology transformation.
Kaila emphasized that Blackstone’s focus on computing finance and ability to offer flexible capital structures from debt to equity sets it apart from traditional venture capitalists. And we are actively investing.
“We are the largest owner of non-hyperscalar compute today in that we can use that compute strategically to support the enterprise, whether it’s neocloud investments or data center investments,” Kaila said.
For example, BXN1 played a key role in several computing-related transactions. These include Crux AI, a neocloud launched in partnership with Google to commercialize custom TPUs, in which Blackstone has invested $5 billion as an investor. It also backed Etched, a four-year-old chip startup whose rapid development has attracted investor attention. Neoclouds is a cloud provider that rents AI GPU infrastructure.
By analyzing committed customer contracts and using GPUs pledged as collateral, Blackstone pioneered a financing structure that allows venture-backed cloud neoclouds to make large loans.
Blackstone’s first such agreement was signed with CoreWeave in 2023, shortly after the launch of ChatGPT led to an explosion in neocloud demand. Kaira said CoreWeave approached Blackstone for a $2.3 billion loan even though the company had “$30 million in revenue and minimal profitability” at the time.
CoreWeave had borrowed a significant amount of its vast collection of Nvidia GPUs, and Nvidia was also an investor. Blackstone also felt more comfortable with financing because CoreWeave had contracts with blue-chip customers such as Microsoft.
The deal quickly became a financing model for Blackstone and other credit institutions to provide billions of dollars in additional debt to Coreweave and rival NeoCloud.
“One of the main things we look at is the nature of the customer. And if that customer is an investment-grade customer, like Microsoft or Meta, the risk is priced accordingly,” Kaila said.
Building on its success with CoreWeave, Blackstone has entered into other neocloud financing deals, including co-leading a $10 billion credit facility with Coatue for Australian GPU provider Firmus earlier this year.
Kaira acknowledged that the huge debt burdens these companies and other neo-cloud operators have accumulated in recent weeks are starting to raise red flags. Since then, Coreweave’s debt alone has ballooned to about $50 billion, with Altimeter Capital founder Brad Gerstner saying in a recent CNBC interview, “The only collateral that CoreWeave is putting up is really the GPUs, but that’s not enough anymore. They’ve used that in the past to raise cheap debt.”
Kaira said that on the credit front, market concerns stem from a combination of factors. Interest rates rose in September and higher-than-expected inflation suggests borrowing costs may continue to rise. The flood of AI infrastructure bond issuances also doesn’t help.
“Credit spreads are also rising,” Kaira said, explaining that lenders are charging higher premiums on top of already high base rates to fund the buildout of AI. “The cost of capital for debt at Meta and Microsoft is increasing,” he added, noting that even tech giants with huge balance sheets and strong profits are paying more to borrow.
Concerns about rising debt costs are also spilling over into the stock market. Coreweave’s share price has fallen more than 20% in recent days, and Farmus has called off a planned IPO on the Australian Stock Exchange with a target valuation of about $30 billion.
“Ultimately, the market didn’t exist at that valuation, so I don’t think that deal went through as a result,” Kaila said of Firmus’ delisting.
“The market is thinking, in this moment when interest rates are rising, what is the right price for neocloud, what is the right price for these businesses,” he said.
Despite this pressure on some infrastructure providers, AI companies just can’t get enough compute to power their models. Their appetite continues to drive data center expansion.
Kaira downplayed concerns from doomsayers that the huge buildup of AI infrastructure debt could lead to a widespread credit crisis, or worse, collapse the entire economy. “This doesn’t look like a 2008 bank with shaky credit,” he said.
Instead, he pointed out that because the majority of data centers are being built for hyperscalers, the financing for those companies is ultimately backed by “the most creditworthy companies in the world.”
In addition to funding data centers, Blackstone is helping companies implement AI through a $1.5 billion joint venture called Ord with Anthropic that it formed this year with Anthropic and Hellman & Friedman. Ode provides forward-deployed engineers to help businesses, including 280 Blackstone portfolio companies, adopt and integrate AI into their operations.
Asked if Blackstone-owned companies are seeing tangible benefits from AI, Kaila acknowledged: “There is definitely an ROI, but it’s variable.” He said technology adoption could proceed in a “sawtooth pattern,” and to skeptics, computing could appear to be in oversupply before adoption catches up.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
