AI Buildout Needs More Equity as Debt Risks Pile Up, Blackstone's Gray Says

Blackstone President Jon Gray said Tuesday that the artificial intelligence infrastructure buildout will need more equity financing, because lenders are becoming dangerously exposed to the companies driving the construction boom.

Speaking at the Hong Kong Monetary Authority's Global Financial Leaders' Investment Summit, Gray said the sheer scale of capital required for AI is creating concentration risks for debt investors — both across the sector as a whole and within individual companies.

The warning matters because the AI buildout is one of the largest capital-raising efforts in corporate history. If the financing structure wobbles, the effects wouldn't stay confined to tech; they would ripple through banks, bond markets and pension funds that have lent into the boom.

Why lenders face a different problem than shareholders

Gray laid out the core mechanism plainly. Shareholders can benefit when a company grows — their upside moves with the business. Lenders can't. Their returns are largely confined to interest payments and the repayment of principal.

That asymmetry becomes a problem when enormous amounts of debt flow into a single sector. If AI infrastructure companies stumble, equity holders lose first, but concentrated debt exposure spreads losses across the financial system.

"There's going to be more equity needed across this space to keep up with these enormous demands," Gray, who is also Blackstone's chief operating officer, said at the summit.

The takeaway: as the numbers grow, the people lending the money have less and less room for error.

How big is the bill? Around $10 trillion

Pimco CEO Emmanuel Roman, speaking on the same panel, put a number on the challenge. AI-related capital spending could reach around $10 trillion over the next five years, he said, citing an estimate from General Atlantic CEO Bill Ford.

Roman added that about $550 billion in AI debt has already been issued in various forms so far this year.

To grasp the scale, Roman offered a comparison: the global corporate bond market stands at roughly $20 trillion. Even with a mix of equity and debt financing, the amount of capital AI demands would be "gigantic," he said.

"At some point in time, it's going to be too much," Roman warned.

His concern is absorption capacity — whether markets can actually take on that much new paper without breaking. When a single sector's financing needs approach half the size of the entire global corporate bond market, the question stops being theoretical.

The one-line version: two of the world's most influential asset managers think the financing side of AI is becoming the bottleneck.

Goldman Sachs sees $8.5 trillion for data centers alone

The warnings from Hong Kong line up with sell-side research. Global data center demand will require about $8.5 trillion in capital spending over the next five years, Goldman Sachs wrote in a report published Monday.

That figure covers just the data center slice of the AI buildout — the physical facilities, power and equipment — not the full range of AI-related investment Roman referenced. Even so, it puts the infrastructure story in the same order of magnitude as the $10 trillion total estimate.

For context, that level of spending rivals major national infrastructure programs, compressed into half a decade and driven largely by private companies.

The takeaway: the numbers from bankers, investors and analysts are converging on the same conclusion — this is a multi-trillion-dollar project with no recent precedent.

Where will the money come from?

Gray sketched the financing mix he expects. Funding the buildout will require a combination of public markets, banks and private capital, he said. Companies are increasingly turning to structured transactions to limit the amount of debt they take on their balance sheets.

Cash generated by the hyperscalers — the giant cloud and AI companies — would also fund part of the spending, Gray added.

"The creativity is needed because the scale of capital is enormous," he said.

Structured deals, which parcel out risk to different classes of investors, are one way to keep plain-vanilla bank debt from piling up. But they also spread exposure in ways that can be hard to trace — a pattern familiar to anyone who watched the pre-2008 credit boom.

The plain summary: the money will come from everywhere at once, and the industry will have to invent new financing structures to move it.

Demand still outruns supply

Despite the financing challenges, Gray stayed bullish on AI infrastructure itself. Demand for computing power is "running well ahead of supply," he said.

He backed that up with a number from Blackstone's own portfolio. Spending on AI models across the firm's portfolio companies has risen 20-fold over the past year, as businesses see financial benefits from applications including coding, customer engagement and content creation.

That 20-fold jump is the strongest argument that the capital rush isn't speculative froth — companies are paying because the tools already pay them back.

Still, meeting that demand is getting harder, not easier. Gray said securing sites, permits, chips, power and capital has become more challenging — a list that covers nearly every input the buildout depends on.

The takeaway in plain words: the customers want more compute than the industry can currently deliver, and the physical and financial constraints are tightening at the same time.

What happens next

The immediate question for markets is who supplies the equity Gray says the sector needs. Public investors, private funds and the hyperscalers' own cash flows will all have to carry more of the load, because lenders — and the regulators watching them — are starting to flag concentration risk.

Roman's warning sets the clock: if AI financing keeps growing at its current pace, markets could hit their absorption limit before the buildout is finished. How the industry restructures its funding in the next few years will determine whether the $10 trillion question gets answered smoothly — or painfully.