During April and May, the Financial Stability Board (FSB), the International Monetary Fund (IMF) and the Group of Thirty (G30) published new assessments that deepened the understanding of vulnerabilities in private credit markets. Against the backdrop of continued market growth - the FSB estimates the narrowly defined direct lending market at USD 1.5–2.0 trillion at the end of 2024 - the reports highlight concerns across four connected areas: the weak credit profile and high leverage of many borrowers; layered leverage and dense interconnections across funds, sponsors and banks; opacity in ratings, valuations and exposures; and the gradual emergence of liquidity mismatch in vehicles offering periodic redemptions.

The first concern relates to the underlying borrowers. Private credit remains concentrated among unrated or lower-rated companies. Where ratings are available, borrowers cluster around single-B and generally carry more leverage than those in the broadly syndicated loan market, with EBITDA add-backs potentially understating effective leverage. Headline payment defaults remain low but are rising from a low base. Broader indicators—including selective defaults and distressed exchanges—show more stress, while increased use of payment-in-kind interest and maturity extensions can defer cash flow pressure rather than resolve it. IMF analysis suggests that the share of distressed direct lending borrowers could more than double under sharply higher interest rates or weaker earnings. The second concern is the accumulation of leverage and exposures across the financing chain. Leverage can arise at the portfolio company, fund and investor levels, with private equity sponsors often adding debt at the acquisition stage. Banks create further links through subscription and portfolio financing facilities, revolving credit lines to common borrowers, securitisation services, synthetic risk transfers and strategic partnerships. Direct bank exposures appear small in aggregate, but estimates vary materially and do not capture all indirect channels. The G30’s assessment is that credit risk has partly migrated to nonbanks, while much of the associated liquidity risk remains concentrated in the banking system, creating several routes through which stress at a borrower or fund could return to banks.

The third concern is opacity. Most borrowers lack public ratings, private ratings and credit estimates can be difficult to compare, and bespoke loans are typically valued quarterly or less frequently using models that allow significant judgement. Divergent marks and delayed loss recognition can obscure deterioration, while inconsistent definitions and limited granular fund and loan level data leave authorities and market participants with only a partial view of common borrowers, leverage and concentrations. Finally, liquidity mismatch remains limited across a market still dominated by closed-ended funds, but is becoming more relevant as evergreen and semiliquid vehicles offer periodic redemptions and attract a broader investor base. The IMF estimates that such vehicles account for around one fifth of direct lending loans. Recent redemption pressure has illustrated the potential interaction between illiquid assets, slow moving valuations and first mover incentives. In prolonged stress, withdrawals could coincide with borrower draws on revolving facilities and difficulties refinancing fund debt, eroding liquidity buffers; gates may preserve fund liquidity but weaken confidence across comparable vehicles, while additional secured borrowing could leave remaining investors with more leveraged exposure. The IMF therefore assesses the current systemic impact as contained, while the FSB identifies scope for the vulnerability to increase as semiliquid and retail-oriented structures expand.

At regional and national level, recent risk assessments broadly point to manageable aggregate exposures. while keeping private credit risk firmly on their radar. In the EU, the European Central Bank finds that limited direct exposures make private credit unlikely to be a standalone source of systemic stress at present. However, the capacity of euro area firms backed by private credit to service interest from operating cash flow has deteriorated, and scenario analysis indicates that broader spillovers to leveraged loans, high yield bonds and equities could generate larger losses than defaults on private credit loans alone. The exposure is also strongly cross-border: euro area investors allocate around 60% of their private credit commitments to foreign funds, while euro area companies receive around 70% of their private credit funding from non-euro area lenders. In Australia, the Australian Prudential Regulation Authority's May System Risk Outlook similarly notes domestic risks as contained, estimating the private credit market at around AUD 200 billion, or approximately 3% of the banking system, with a primary concentration in real estate. It identifies cross-border transmission as the more relevant risk: around 16% of superannuation fund investments are in private market assets, approximately half of which are offshore, while banks have increased their appetite for international funds finance. In Canada, the Office of the Superintendent of Financial Institutions (OSFI) elevated non-bank financial institution (NBFI) risk to a top risk as part of its Annual Risk Outlook released in April.

Insurers, particularly life insurers, are becoming increasingly important sources of capital for private credit. The attraction reflects both return and asset-liability management considerations: long-dated private loans can provide stable cash flows, an illiquidity premium and duration matching for annuity and other long-term liabilities. Exposure may rise further as private credit origination broadens and alternative asset managers expand their links with insurers through ownership, affiliated asset management and funded reinsurance arrangements.

The FSB’s May report identifies several insurer-specific channels through which this trend could create vulnerabilities. Life insurers hold higher private credit exposures than non-life insurers, while private equity ownership and funded reinsurance are strengthening links between insurance balance sheets and alternative asset managers. The FSB notes that these arrangements can create complex and opaque structures, potential conflicts of interest and difficult-to-detect pockets of risk. Separately, private ratings may facilitate insurer investment in private credit and affect regulatory capital treatment.

Recent EIOPA releases provide further detail on the scale and composition of these exposures. In its June Financial Stability Report, EIOPA estimated EEA insurers’ private credit exposure at EUR 523 billion at year-end 2025, equivalent to 5.0% of total assets. Private credit represented 12.8% of life insurers’ general account investments, compared with 6.2% for non-life insurers, 3.3% for reinsurers and around 1% for unit-linked portfolios. In a subsequent dedicated factsheet, EIOPA showed that 68.9% of the exposure comprised mortgages and loans and 21.1% non-listed or non-traded corporate bonds. These figures use a broad supervisory definition that extends beyond sponsor-backed direct lending; EIOPA notes that mortgage exposures can have materially different risk characteristics from lending to highly leveraged firms. The FSB cites a North American proxy based on private placements and private ratings that places private credit at around 10% of life insurers’ portfolios, compared with around 3% for non-life insurers.

Recent assessments of private credit devote specific attention to its interaction with AI as one element of the market’s evolving risk profile. Notably, analysis by the FSB and several deep dives by the Bank for International Settlements (BIS) identify two distinct channels linking private credit and AI: the financing of new AI infrastructure and the exposure of existing borrowers to AI-driven disruption. In its report on private credit vulnerabilities, the FSB cites private sector projections of USD 2.9 trillion in AI infrastructure capital expenditure during 2025–28, of which USD 1.5 trillion would be externally financed and USD 800 billion supplied by private credit. Separately, BIS authors using PitchBook data estimate that outstanding direct loans to firms classified in the "Artificial Intelligence", "Big Data" and "Cloud Tech" verticals exceeded USD 200 billion by 2025, equivalent to almost 8% of total outstanding direct loan volume in that dataset. Participation has broadened, but fund-level exposure remains modest: around 20% of private credit funds have lent to AI-related sectors, and such loans account for about 5% of the average fund’s loan volume.

The infrastructure channel is most visible in data centre finance. Corporate bonds remain the main source of debt funding for large technology companies, but private credit is gaining ground in large, asset heavy projects. Bespoke, asset-based structures can match long-dated project costs to contracted lease or capacity payments. Yet the same flexibility can shift leverage into complex off-balance sheet arrangements. Technology companies increasingly use joint ventures or special purpose vehicles to acquire or develop data centres, typically retaining a minority equity stake and supporting the vehicle through long-term leases, capacity commitments and, in some cases, guarantees. Private credit funds and other institutional investors, including insurers, hold the vehicle debt, while banks may provide funding lines. These structures replace upfront capital expenditure with multi-year commitments and leave much of the associated borrowing outside the technology company’s balance sheet. Stress can therefore transmit through vehicle level refinancing pressure, a pullback in private credit or the activation of guarantees. Credit performance remains sensitive to power availability, construction delays, tenant concentration, overcapacity and the long-term value of data centre collateral. AI-related private loans are also materially larger than other loans on average—USD 169 million compared with USD 90 million—despite similar maturities and spreads. The BIS notes that the gap between similar debt pricing and elevated equity valuations points either to underpriced credit risk or overly optimistic expectations for future AI cash flows.

The second channel runs through existing borrowers whose business models are exposed to AI disruption, most notably software-as-a-service companies. Outstanding direct loans to these firms climbed from around USD 8 billion in 2015 to more than USD 500 billion—19% of total direct loans—by as at the end of 2025, and roughly one third of private credit funds had lent to the sector. Publicly traded business development companies provide a clearer window into how investors are pricing that exposure. Between October 2025 and February 2026, software equities fell by almost 30%; by early March, Business Development Company (BDC) shares were down about 10%, and software-heavy BDCs had underperformed lower exposure peers by roughly 5 percentage points. Wider discounts to net asset value pointed to concerns about the carrying value of underlying illiquid loans. Taken together, the BIS analyses highlight a two-sided risk: newly financed AI infrastructure may fall short of expected returns, while AI adoption may weaken the business models and credit profiles of existing technology borrowers. The broader features of private credit—sector concentration, limited transparency and infrequent valuations—could amplify both channels, particularly in redemption-enabled vehicles during periods of withdrawal pressure.