AI Infrastructure Bets Will Show Up in 2027 Performance Fees

What's New

Blackstone's AI deployment is about to turn into realized performance fees and faster management fee growth. Michael Chae, CFO at Blackstone, argues this in a fireside chat at the Barclays 24th Annual Global Financial Services Conference. He points to data centers, power, and stakes in frontier AI companies as the source of accrued carry across several vehicles. The first large test is a scheduled crystallization in the institutional infrastructure strategy in the fourth quarter of 2027. Allocators should track realized fees from AI assets and discount accrued ones until they pay out.

Why It Matters

Investors who see AI spending as a cost with a distant payoff are on the other side of this argument. Chae says the path from deployment to earnings is already visible in accrued performance revenue. He also challenges the view that managers need a reopened sponsor M&A market to realize gains. Blackstone calls itself the largest owner and developer of data centers in the world, so the thesis supports its own fee outlook.

What To Watch

• The fourth-quarter 2027 crystallization in the institutional infrastructure strategy. Accrued performance revenue there was over $900 million in the second quarter.

• Incentive fees from BXPE. Its NAV has reached $27 billion, and Chae ties part of its performance to exposure to AI companies.

• Pricing on the 6 IPOs Blackstone has on file. Chae says IPO markets favor large, high-quality companies, which is how he expects accrued carry to become liquid.

Reality Check

Accrued performance revenue is a mark, and Blackstone sets it. Chae gives no realized figure for the frontier AI stakes, only that they have had "exceptional upside." He concedes internal AI use is early. He also expects private credit defaults to rise from historic lows. The infrastructure crystallization is more than a year away.

Memorable Quotes

• "Despite the scale of the AI build-out, these investments so far remain less than 1% of GDP." Chae uses the ratio to argue the cycle has room to run.

• "The key question on selling assets is really, are you exposed to the places the world wants?" In his framing, exits depend on sector exposure.

• "I know you all may feel like Steve, Jon, and I have been talking at you every quarter about our AI deployment." The admission sets up the test: the talk now has to show up in earnings.

• "Deployment leads to performance, and performance to performance fees, and also management fee growth." This is the full thesis in one line.

The Wrap

AI-linked assets need to hold their value through realization, and buyers need to keep paying for large, liquid IPOs. Stable data center lease terms and steady long-end rates keep the case intact. Falling frontier AI valuations before exit would break it. Chae's guidance calls for double-digit growth in base management fees next year. The infrastructure crystallization at the end of 2027 will settle whether accrued AI gains become cash.