U.S. private credit, caught in the AI hype bubble

Ask AI · How AI Technology Is Shaking Up SaaS Business Models and Triggering a Private Credit Crisis?

© Read Finance / Produced

Author: Yang Yang

Editor: Xia Yijun

When Middle East artillery fire tugs at the sensitive nerves of global capital markets, the private credit market in the United States—across the ocean—is also hiding a financial risk that cannot be ignored beneath the surface.

In the recent earnings season, the performance of private-credit giants is nothing short of strong. Last year, Blackstone’s net profit was $2.14 billion, doubling year over year and delivering the “strongest performance in 40 years”; last year, KKR raised $43 billion in Q4, a four-year high.

But behind the impressive results, the capital markets have been in a relentless slide. Since the start of the year, Blackstone is down more than 32%, while KKR is down more than 38%.

With performance diverging from stock prices, the core issue is that crisis signals in U.S. private credit are emerging in a concentrated and forceful way.

In the first quarter, BlackRock announced redemption restrictions for its private credit funds, triggering a liquidity “lock-up.” Even Blackstone’s private credit funds prompted employees to reach into their own pockets to subscribe $150 million in order to handle redemptions that exceeded the limit.

The wave of redemptions in private credit is not driven by a fundamental deterioration in credit assets, but by investors’ concerns about AI black swan risks.

Around 40% of the capital in private credit flows to software companies that lack physical collateral, but have long-term steady growth in annual recurring revenue (ARR). For a considerable time in the past, these software companies were among the best yielding assets in the credit market: ARR guarantees cash flow and is considered a high-safety asset, but because they lack collateral, it is difficult to obtain bank lending. Private credit can lend at a high interest rate of 10%-14%.

But AI has shaken this logic. Investors are worried about AI’s impact on the SaaS software business model—just as the market fears the full deployment of intelligent agents could deprive the SaaS industry of its survival foundation.

Once belief is shaken, the chain reaction is already visible: the valuation of credit assets continues to move downward, raising market risk expectations; rising risk further triggers investor redemptions, forcing private institutions to tighten lending conditions; and tightening credit conditions directly puts pressure on software companies’ cash flow, which in turn further suppresses asset valuations, ultimately forming a vicious cycle.

In the future, will private credit evolve into the next financial crisis?

/ 01 /

Private credit gets “locked up”

The end of one crisis often plants the seeds for the next.

In 2008, after the U.S. reviewed the subprime mortgage crisis, the core conclusion was: the “badness” was baked into the loans from the start—loans were made to people who couldn’t repay at all. Therefore, the regulatory logic was: as long as loan quality is good, the system will be stable.

Under this logic, the Basel III Accord and the Dodd-Frank Act were introduced in succession. They restrict banks’ high-risk businesses, making banks more inclined to favor traditional, stable business areas—such as relatively steady sectors, large enterprises, and low-risk customers.

However, on the other side of tightened bank risk controls is that the U.S. economy contains a large number of small and medium-sized enterprises and innovative companies. They are numerous, but because of lower credit ratings, insufficient collateral, or inability to meet bank risk criteria due to innovative business models, they fall outside what banks can approve.

As a result, financing that originally came from banks gradually shifted to non-bank institutions, and private credit began to take on the financing supply that banks were shrinking away from.

So-called private credit is, relative to bank credit, debt financing provided directly by non-bank institutions such as private companies and Business Development Companies (BDCs) to enterprises, with no public trading.

The asset management size of private credit has grown from about $200 billion in 2015 to $2.3 trillion in 2025, accounting for 30% of the total debt financing size for below-investment-grade enterprises in the United States.

For a long time, private credit has been promoted by Wall Street as a “safe haven” that offers stable returns for individual investors.

One reason is that the underlying quality of private credit is not bad. About 40% of the funds in private credit flow to software companies.

As asset-light companies, software firms lack physical collateral, making it hard to secure loans through traditional channels like banks. Therefore, private credit provides them with high interest rates of 10%-14%. But SaaS software’s annual recurring revenue (ARR) has long continued to grow steadily, and they have reliable debt-repayment capacity.

This product upgrade, which seems to offer arbitrage opportunities, ran into a redemption wave in the first quarter—then even triggered a major-institution “lock-up”:

BlackRock’s corporate loan fund (HLEND) saw a record 9.3% redemption request. Management was forced to lock the buyback cap at 5%, with roughly $1.2 billion in redemption requests “paused.”

Blackstone’s private credit fund BCRED reached a record 7.9% in redemption requests, exceeding the 7% cap. To avoid triggering the lock-up mechanism, Blackstone required employees to subscribe $150 million out of their own pockets.

Blue Owl, to address the liquidity crisis caused by redemptions, has even already sold $1.4 billion worth of credit assets.

What crisis has private credit run into?

/ 02 /

Caught by the AI black swan

In recent times, AI agents represented by Anthropic have become killers of SaaS.

The papers or tools released by Anthropic have already caused a 20% single-day plunge in Salesforce, and a 13.15% single-day drop in IBM’s stock…

And since software accounts for about 40% of the share in private credit, the redemption wave in private credit can be seen as an extension of the sharp drop in U.S.-listed SaaS software.

Specifically, the redemption wave among investors is not driven by a significant deterioration in credit assets. In fact, software bellwethers like Salesforce and IBM achieved both revenue growth and profit growth in their latest quarterly reports.

Investors’ main worry is the impact of AI on SaaS software business models—such as market fears that intelligent agents’ full deployment could remove the foundation for the SaaS industry to survive.

The logic is that SaaS’s business model is built on a “software-as-a-service” subscription model. Its core charging logic is licensing by user seats (Per Seat) or by functional modules. Enterprises essentially “rent” standardized software tools for employees—paying, in essence, for functions.

But AI agents interact directly with backend systems via APIs, without requiring users to operate complex interfaces, and can bypass SaaS usage scenarios to pay directly for results. Once the SaaS business model wobbles, the money in private credit may no longer be as stable.

An overseas bearish article, titled “The Global AI Crisis 2028,” raised concerns about this. It also assumed a core case:

Zendesk (a leading U.S. customer-service SaaS company) was taken private in 2022 for $10.2 billion, securing a $5 billion direct loan collateralized by ARR. But if by 2027 AI customer service fully replaces the human ticketing system, Zendesk’s core business logic would disappear, and the $5 billion loan could very possibly go to waste.

Of course, some investors may say private credit can invest in AI. But the reality is that AI is currently a bit “above being beyond criticism”: huge investments have been poured in, yet they have not created matching revenue. For example, Open AI has just raised $122 billion, but its annualized revenue is currently only $20 billion. Investors also have concerns about an AI bubble.

For instance, in December 2025, private credit giant Blue Owl Capital exited a $10 billion financing project for Oracle’s AI data centers. The market interpreted this as private credit institutions becoming more cautious in pricing the risks of AI investments, starting to question whether tech giants’ aggressive expansion via off-balance-sheet financing is sustainable.

Actually, private credit originally is a “long money, long loans” product—providing long-term financing to borrowing companies—so in theory it is not afraid of runs. But to make the borrowing side more attractive to customers, private credit funds often structure themselves with “lock-up periods + quarterly/semi-annual redemptions + redemption caps,” meeting investors’ psychological expectations of having an “exit.” With current concerns about AI risks, this design has led to the birth of the redemption wave.

Indeed, fate has already set the price for the gifts it gives you in the dark.

/ 03 /

U.S. software companies caught in a vicious cycle

The 2008 financial crisis is still fresh in people’s minds. Will the redemption risk in private credit further expand this time and affect the U.S. financial system, bringing about a new crisis?

At present, most institutional views believe that U.S. private credit is still more of a liquidity and trust crisis, rather than a full-scale solvency crisis.

Goldman Sachs, using a stress-testing framework for private credit default scenarios, pointed out that even in an extreme scenario where the default rate rises to 10%, the drag on GDP would be only 0.2% to 0.5%. Under the baseline scenario, private credit default rates would rise from about 1% in 2025 to 3% to 4%—which corresponds to the low-end range of leveraged loan default rates in historical credit cycles—so overall risk remains within a controllable range.

In terms of scale, private credit is currently about $2.3 trillion, and its leverage ratio is not high in itself (around 1x). By comparison, the commercial real estate loan scale that worried the market in 2022-23 is about $4 trillion, and in 2008, subprime mortgages plus derivatives totaled on the order of tens of trillions of dollars.

Although the risk of U.S. private credit evolving into a financial crisis is relatively small, U.S. software stocks could still see valuation declines due to a private credit crisis—leading to a vicious cycle: private credit faces pressure on its balance sheet as assets and liabilities strain, credit conditions tighten, and then software companies’ cash flows worsen.

Transmission begins with lowered valuations of software companies, amplifying investors’ risk. For example, if software stocks get cut in half due to AI’s disruption of business models, loans extended against those stocks—originally valued at $10 billion—may leave collateral worth only $5 billion, magnifying investors’ risk.

As risks rise, private credit funds will face larger redemption waves. At this point, private credit can only sell the assets that are easiest to sell to raise cash, and will tighten lending conditions even more. The direct result is that financing for software companies suddenly becomes much harder.

Taking into account that many U.S. small and mid-sized SaaS companies have relied on private credit to survive—even if they are not profitable, as long as revenue grows and valuations stay high—they will, once funds start doubting that these companies can survive being replaced by AI, be unwilling to extend new loans or will demand higher interest rates and lower valuations. The outcome is that the software industry will first experience a wave of layoffs, a financing freeze, and a wave of bankruptcies.

Private credit may not necessarily trigger a large-scale financial crisis, but the impact on assets that could be eliminated by the times should not be underestimated.

Disclaimer: This article (report) is written based on publicly available information or the information provided by interviewees. Read Finance and the author do not guarantee the completeness or accuracy of such information. Under no circumstances does the information or the opinions expressed in this article (report) constitute investment advice to any person.