Blackstone's Gray Says AI Demand Is 'Almost Parabolic' as Supply, Not Demand, Is the Constraint

Gray described the quarter as challenging. Energy prices rose about 30% and interest rates rose about 80 basis points, and those moves slowed transaction activity somewhat.

Even so, Gray said Blackstone delivered for clients across the board, and that its flagship strategies beat comparable public market indices, which he said is what the firm is really focused on. The firm raised $63 billion of inflows this quarter and more than $270 billion over the last 12 months, the most in four years. Fee-related earnings rose 6% in the quarter and overall earnings are up 12% year to date, although realizations were somewhat lumpy.

Burger noted there is no end in sight to the war in Iran and that the Fed is not discussing rate cuts, and asked what would need to change to realize the strength Gray sees ahead. Gray said the firm is assuming it will be in a choppy world for a while and is not simply planning for conditions to get much better.

He said a resolution of either or both of the conflicts in Russia-Ukraine or the Middle East would be very helpful, easing pressure on the Fed and other central banks and bringing inflationary pressure down. He also argued that people are overlooking underlying economic strength, particularly in the United States, with private equity companies seeing around 10% revenue growth and continued margin expansion.

Burger asked whether Gray sees bubbles in AI, noting that Ken Griffin had seen signs of them and that Ray Dalio has been watching for one. Gray said he sees it differently. A bubble, in his view, means building capacity and hoping demand arrives, while Blackstone's portfolio companies are seeing demand that already exists. He compared the situation to the 1990s telecom fiber bubble, when companies built large amounts of fiber on heavy leverage and the outcome was poor.

Gray said demand for model spending at Blackstone's portfolio companies is up 20-fold year on year. Blackstone also created a company with Anthropic called Ode to deploy AI, which he said cannot keep up with customer demand, and personalized agents are adding incremental demand as well. He concluded that demand is almost parabolic.

On the supply side, Gray said the constraints are entitlements, power, ships, and the capital needed to assemble them. Blackstone is focused on long-term energy assets and data centers contracted to the largest companies. He said there is an indisputable shortage of compute that is hard to deliver, and that this shortage is what prevents the kind of excess supply of capital seen in past bubbles.

Burger pointed to a company Gray has worked with on ship financing whose debt yields have reached 15%, and asked whether the broader rise in rates is adding to the cost of capital for the AI buildout. Gray said higher rates are creating a challenge.

He contrasted the supply side of delivering compute with demand for it, which he described as incredibly strong, and used neo clouds as the example.

Gray said that for a Neo Cloud able to deliver compute, the price it can get today is up 50% from a year ago. Older chips, such as the 8181 hundreds, are releasing close to their original rates, even though people thought three or four years ago they would have no value.

He also said lease rates on the data centers contracted have risen, so existing leases are now in the money. As a result, he said the main business for Neo Clouds is quite good.

Gray said the challenge for Neo Clouds is delivering the compute and honoring their commitments, which requires amassing debt, equity, data center capacity, and power. Failing to do so is what creates the challenge.

He said CoreWeave has an outstanding CEO who is very creative and will likely find a way through, but that navigating the current conditions is the real market test today.

Gray said that as long as compute demand and pricing power hold, Neo Clouds will find sources of capital. They can do chip financings at the asset level and may have to raise equity, and each company will face these issues differently. He said a weaker end market would be highly problematic.

Gray said Blackstone's strength is its access to large amounts of private capital, allowing it to act as a solutions provider: investment-grade credit at lower cost, more junior credit at higher cost, participating equity, and common equity. He expects the buildout to be financed through a mix of all of these.

Asked whether an Anthropic IPO going smoothly could help calm the market, Gray said it would be a real positive. He pointed to SpaceX, noting that when large private companies go public they get big market caps, and rating agencies often make them investment grade, which lowers their cost of capital.

He said the big labs have had to rely on hyperscalers for data center leasing and chip financing, and that their own market cap and credit rating would let them access capital more easily and cheaply. Gray said this would reduce pressure on the system, and that more capital entering the system would be helpful.

Gray said at the beginning of the year that 2026 would be the year of the IPO, and on paper that is coming to pass year to date, with US IPO volume up about 3.5 times. If Anthropic completes its offering, total volume would rival 2021. Five companies have gone public so far, three in the US and two in India.

Gray acknowledged that choppiness in markets has delayed some IPOs and narrowed the window. He said greater stability in the geopolitical picture is needed before conditions improve, but that the IPO window will reopen once real earnings strength shows through in public and private markets.