The LP Letter That Cannot Answer the AI Question: Private Credit’s Disclosure Crisis

Ask a private credit limited partner how much of their allocation is exposed to AI-displacement risk in software borrowers, and the honest answer is that they do not know. Their fund letter tells them what percentage of the portfolio is in “software.” It does not tell them how that software exposure breaks down between infrastructure, vertical SaaS, and horizontal application software—the three categories that carry very different AI-substitution risk profiles. That gap is directly generating the redemption wave now running through the asset class.

Disclosure Was Adequate for Generic Risk. It Is Not Adequate for This Risk.

Private credit disclosure standards evolved to address the risks that were visible and broadly understood: leverage levels, interest coverage ratios, covenant compliance rates, sector concentration. These metrics allow LPs to assess whether a portfolio is stressed in the ways that credit portfolios have historically been stressed—economic slowdown, commodity price cycles, rate moves.

AI-displacement risk does not fit any of those categories. It is a structural question about whether specific software business models remain viable as AI capabilities develop. It affects different sub-segments of the software market on different timelines. A fund that discloses aggregate software exposure at 28% of AUM is giving an LP almost no information about whether that 28% is in infrastructure code (largely defensible) or horizontal productivity software (high AI-substitution risk). The disclosure standard was built for a different risk environment.

The Structural Chain Eileen Appelbaum Identified

CEPR co-director Eileen Appelbaum’s April 2026 analysis documented the chain of decisions that built the current exposure: PE firms acquired life-insurance and annuity businesses over seven years, redirecting policyholder reserves into proprietary private credit funds with minimal disclosure and no mark-to-market discipline. Those funds lent extensively to PE-owned software companies, particularly between 2022 and 2024, when the software lending thesis was at peak confidence and leverage multiples ran at six to eight times EBITDA.

The insulation the structure provides from market feedback—the same feature that made it attractive—is now the problem. Marks do not move until the fund manager updates them. Disclosure does not expand until the LP base demands it loudly enough. By the time the data reflects deteriorating conditions, the conditions have typically been developing for several quarters.

Gates at Three Funds Since March

Two of the largest perpetual private credit vehicles imposed quarterly outflow caps in March 2026. A third followed in April. None disclosed material credit losses alongside the announcements. Secondary buyers of fund interests have priced discounts against stated NAVs—assigning probability to future mark-downs that the fund’s quarterly valuation process has not yet reflected. Each gate announcement has generated additional redemption pressure from LPs who read the cap as a signal to act before conditions tighten.

Which Portfolios Are Most Exposed

The risk is concentrated in portfolios that lent heavily to horizontal application software between 2022 and 2024. These books carry the highest density of borrowers whose revenue models face near-term AI-substitution pressure. Portfolios built around infrastructure software, deep-vertical SaaS with regulatory and workflow dependencies, or asset-backed lending outside of software are facing a different conversation with their LP bases.

The manager community argues that private direct lending’s structural advantages—covenant quality, private workout mechanics, no forced-sale environment—differentiate it from public high-yield stress cycles of 2008 and 2015. Those structural advantages are real. They describe how a stress scenario would be managed, not what it would cost. The cost depends on the revenue trajectory of software borrowers through 2028, which is not yet known and not disclosed by the fund structures.

NAV prints over the next two quarters are the first hard evidence point. The appearance of AI-displacement-risk metrics in LP letters—a development that requires sustained LP pressure to produce—would signal that the disclosure standard is beginning to catch up with the risk environment. Both data points are quarters away. The redemption queue is the most current evidence the market has about how LPs are voting with their capital.

Source: Private Credit Fund Redemptions Climb Sharply, Some Caps Now in Place