Blackstone’s Jon Gray calls for more equity as AI lending risks grow

AI’s infrastructure boom needs more investors willing to own a stake, not just lend money. Blackstone President Jon Gray said November 3 that more equity financing will be needed as lenders’ exposure to the sector grows. His argument pairs confidence in computing demand with a warning about who can keep funding it.

Speaking at the Hong Kong Monetary Authority’s Global Financial Leaders’ Investment Summit, Gray described risks building across both the AI sector and individual borrowers. On the same panel, PIMCO CEO Emmanuel Roman raised a broader concern: the spending could become too large for markets to absorb, CNBC reported.

Lenders’ exposure, shareholders’ upside

Gray’s concern centers on concentration: debt investors can accumulate too much exposure to the same industry or company. He contrasted their position with that of shareholders, who can benefit from a company’s growth. Lenders’ returns, by comparison, are largely limited to interest and repayment of the money they advanced.

That distinction underpins his call for more ownership capital rather than simply more borrowing. His proposed funding mix still includes public markets, banks and private capital. Cash generated by hyperscalers—the large cloud-computing companies—would also pay for part of the buildout.

Companies are increasingly using structured transactions to limit how much debt they take on, Gray said. He presented those arrangements as part of a financing effort that will need several sources of capital, rather than one channel capable of meeting the entire bill.

The creativity is needed because the scale of capital is enormous,

Jon Gray, Blackstone president, speaking at the Hong Kong summit, as quoted by CNBC

Roman questions the total bill

Roman’s warning went beyond choosing between lenders and shareholders. Citing General Atlantic CEO Bill Ford’s estimate, he said AI-related capital spending could reach about $10 trillion over the next five years. Even combining debt and equity would leave a gigantic financing requirement, he argued, comparing it with a roughly $20 trillion global corporate bond market.

He said about $550 billion in AI debt had already been issued in various forms during 2026. That figure describes financing raised so far, while the five-year spending estimate describes a possible future requirement. Roman’s conclusion was blunt: at some point, the amount could be too much.

Goldman Sachs offered a more constructive view in its October 28 financing analysis. Mahesh Saireddy, global co-head of its Capital Solutions Group, said he believed the buildout would get financed, while warning of market corrections. Goldman also noted that the largest cloud companies remain relatively lightly indebted and generate strongly positive free cash flow.

Strong demand, recurring costs

Gray’s financing warning did not signal a retreat from his demand outlook. He said demand for computing power was running well ahead of supply. The obstacles extend beyond money: securing sites, permits, chips and power is also becoming more difficult, he said.

As evidence of business adoption, Gray said spending on AI models across Blackstone’s portfolio companies had risen 20-fold over the past year. He attributed that increase to financial benefits in areas including coding, customer engagement and content creation. The figure measures spending within Blackstone’s portfolio, not industry-wide adoption.

Goldman’s analysis adds a longer-term complication: much of the spending goes toward equipment inside data centers that needs more frequent replacement. Saireddy described recurring investment cycles as technology evolves, rather than a single construction bill. Its financing outlook therefore rests on matching capital to assets with different lifetimes and risks—not merely raising one enormous pool of money.