The rise of artificial intelligence is forcing investors to reassess risks across the $2T private credit market, and commercial real estate debt is caught in the crossfire.

The private credit universe is facing repricing pain as Wall Street tries to divine the winners and losers of the rise of AI, with concern over software and corporate loans rising while loans backed by real assets like data centers and offices largely perform as predicted.

At BridgeInvest, a private capital senior mortgage lender in the process of raising a $1B multisector real estate fund, management has recently spent more time explaining the difference between corporate debt and debt backed by real estate to potential investors.

“We spend some time differentiating ourselves from that side of the private credit world,” said BridgeInvest General Counsel Isaac Marcushamer. “Yes, you can buy a car, but there is a difference between a Lamborghini and a Ferrari.”

Private credit stocks have been hammered in recent weeks alongside real estate brokerages and other service-related companies as the so-called AI scare trade takes hold.

The fears aren’t entirely unwarranted: Financial giant BlackRock this month cut the value of its roughly $25M loan to e-commerce firm Infinite Commerce Holdings to zero, the second recent total wipeout in its private credit division.

Warning signs around private credit emerged last year with the bankruptcies of First Brands and Tricolor, both automotive lenders. Those collapses prompted JPMorgan Chase CEO Jamie Dimon to warn of more private credit-related bankruptcies, declaring in October that "when you see one cockroach, there are probably more."

But the public market worries over private debt have been focused largely on loans provided to software firms — or to companies with apparent exposure to AI disruption — where repayment prospects are dependent on the company’s success.

Real estate debt, even for AI infrastructure like data centers, is backed by the physical asset, adding a store of value that the lender can claw back if the loan goes bad.

Private capital exploded onto the scene as a source of AI investment over the past three years, growing from practically nothing in 2022 to more than $200B today, according to a January report from the Bank for International Settlements.

Debt funds had 13% of the exposure to the $4.8T total U.S. commercial real estate mortgage market at the end of 2024, compared to a 9% average from 2015 to 2019, according to Principal Asset Management.

Loans to AI-related companies have grown to 8% of outstanding private credit debt but remain modest for the average private credit fund. Roughly 20% of all funds in the private credit sector invest in AI-related sectors today, up from 5% in 2010, according to BIS.

There’s no shortage of capital looking for deals, and relatively strong demand for debt backed by data centers or apartment complexes have pulled down margins and created room for banks to push back into the marketplace.

“Spreads have come in, and there's more competition and there's more deals,” said Ran Eliasaf, founder of Northwind Group, a private real estate lender focused on multifamily and healthcare assets across 25 states. “You need to be more selective, and you'll only realize the winners and losers two or three years from now.”

The growing reliance on debt to fund AI introduces vulnerabilities to the broader financial system that could have spillover effects, “not least given the rapid growth of less transparent private credit markets and circular financing within the AI ecosystem,” the BIS report notes.

At the same time, the international organization made up of central banks estimates outstanding private credit at AI firms could more than double to up to $600B by 2030.

“Private credit got a little unlucky with some of the challenges that came up for some lenders at around the same time that AI adoption accelerated,” Marcushamer said. “It’s more just headlines being next to each other — I don’t know that private credit is in any worse or better of a position than any other entity.”

The narratives rattling traders are focused on private credit’s exposure to AI broadly, with real estate debt less of a concern, with a recent notable exception. Blue Owl Capital, a publicly traded lender facing pressure for its AI investments, saw its stock wobble after Business Insider reported that it had struggled to shop debt for a CoreWeave-tenanted data center in Lancaster, Pennsylvania.

Blue Owl denied the reports and denied any financing challenges. Around the same time, it sold a $1.4B portfolio of loans in February to generate liquidity for investors exiting another fund, with the buyer group including a Blue Owl-owned insurance firm.

When it comes to real estate, private credit and the broader debt markets are more focused on ensuring that new capital allocations will be on the right side of AI disruption.

Trepp tracked a widening in balance sheet lending spreads for lower-leverage office loans in February amid the stock sell-off, potentially signaling a repricing of the chances for AI to empower companies to cut headcounts and footprints.

“Nothing is fully insulated, ever,” Eliasaf said, pointing to a prediction from Anthropic CEO Dario Amodei that unemployment could reach up to 20% as AI is widely adopted.

“Of course, it's going to hit real estate, there's no doubt, and real estate will adapt. One of the reasons we focus on residential is because the one thing people always need is somewhere to sleep,” he said.

The shorter terms typically offered by private real estate lenders compared to bank or CMBS loans provide some protection against major shifts in asset value. That relative flexibility is increasingly attractive as conflict spreads across the Middle East, tariff policy whipsaws, and the independence of the Federal Reserve is questioned.

Office underwriting is especially sensitive, with elastic demand forecasts that attempt to balance what’s expected to be widespread displacement of some roles — administrative positions, service-oriented jobs and writers are ideal candidates for digitization — with as-yet undefined new professions that support the budding AI ecosystem.

But the same technological leaps that could make that office space obsolete will eventually show up in other asset classes, Marcushamer said. An apartment building that’s mostly occupied by programmers and engineers at a software firm ripe for disruption might not be a good candidate for a loan today, he said.

“There’s no magic line item for AI in underwriting,” he said. “It’s why we pay close attention to everything.”