AI Loan Leverage Hits $1.2 Trillion, Raising 'Credit Termites' Risk

The collateralization of artificial intelligence models and related intellectual property has created over $1.2 trillion in hidden use within the private credit market, a development now being scrutinized as a systemic risk. MarketWatch reported on 23 May 2026 that this opaque use, termed 'credit termites', presents a greater threat than traditional credit-market issues highlighted by JP Morgan Chase CEO Jamie Dimon. The concentration of these loans within mid-tier investment funds has grown 40% year-over-year, far outpacing conventional corporate lending growth.

Context — [why this matters now]

The current risk parallels the structured investment vehicle (SIV) failures preceding the 2008 financial crisis, where $400 billion in off-balance-sheet assets triggered a liquidity freeze. Today's macro backdrop features the 10-year Treasury yield at 4.31% and the Federal Reserve holding its policy rate steady, compressing traditional lending margins and pushing investors toward higher-yielding, complex private assets. The catalyst for heightened focus is the convergence of three factors: a sharp 15% year-to-date decline in AI chipmaker valuations, new regulatory scrutiny from the European Banking Authority on digital asset collateral, and the first quarterly defaults in private credit since 2023, totaling $4.7 billion in Q1 2026.

Data — [what the numbers show]

The estimated $1.2 trillion AI loan market represents 18% of the total $6.7 trillion private credit universe. Leverage ratios on these loans average 6.5x EBITDA, compared to 4.2x for standard leveraged loans. Data from the Loan Syndications and Trading Association shows spreads on AI-backed loans compressed to 375 basis points over SOFR in late 2025, but have since widened to 550 bps as of May 2026. A sector comparison reveals stark differences: the AI loan market's annualized default rate has risen to 2.1%, while the high-yield corporate bond index default rate is 1.4%. The rapid growth is visualized in the change from year-end 2024 to Q1 2026.

Metric	Q4 2024	Q1 2026	Change
AI Loan Market Size	$850B	$1.2T	+41%
Avg. Loan-to-Value Ratio	68%	74%	+6pp
Number of Originating Funds	47	89	+89%

Analysis — [what it means for markets / sectors / tickers]

Second-order effects will pressure specific sectors and tickers. Publicly traded business development companies (BDCs) with high AI exposure, like PSEC and FSK, face immediate mark-to-market losses estimated at 8-12% of their portfolio value. Conversely, large custodial banks and trust providers, including State Street (STT) and BNY Mellon (BK), may see increased fee revenue from acting as independent verifiers of AI collateral. The primary counter-argument is that AI model revenue generation is fundamentally different from physical asset depreciation, potentially justifying higher collateral values. However, this view is challenged by rapid technological obsolescence. Positioning data shows hedge funds, including Citadel and Millennium, have increased short positions in the iShares iBoxx $ High Yield Corporate Bond ETF (HYG) by $3.2 billion notional over the past month, while rotating long exposure into short-duration Treasury ETFs like SGOV.

Outlook — [what to watch next]

Immediate catalysts include the Financial Stability Oversight Council's systemic risk report due 15 June 2026 and earnings from major private credit managers Blackstone (BX) and Ares Management (ARES) on 24 July. Market participants will watch for any disclosure of increased provisions for AI loan losses. Key technical levels to monitor are the 4.50% yield on the 10-year Treasury, a breach of which could trigger further risk-asset repricing, and the 95 support level for the HYG ETF. If the SEC finalizes its proposed rule on enhanced private fund reporting by Q3, forced transparency could trigger a swift deleveraging event across the sector.

Frequently Asked Questions

How do AI loans differ from traditional asset-backed securities?

AI loans are secured by intangible assets like proprietary algorithms, trained datasets, and software code, unlike ABS backed by physical receivables. Valuation relies on projected future cash flows from AI services, not historical payment data. This makes them highly sensitive to shifts in technology adoption rates and vulnerable to model cannibalization by newer, open-source alternatives, creating a faster depreciation cycle than auto loans or credit card debt.

What is the direct risk to a typical investor's bond fund?

Many institutional-grade corporate bond and diversified income funds have allocation limits to private credit, often up to 15%. A 20% write-down in the AI loan segment could translate to a 1-3% direct NAV impact for these funds. The greater risk is contagion: a fire sale in private credit could freeze liquidity for all below-investment-grade debt, widening corporate bond spreads across the board and triggering broad-based ETF outflows.

Which regulators are most focused on this risk, and what tools do they have?

The European Banking Authority is leading with proposed digital collateral frameworks, while the U.S. SEC is focused on disclosure. Their primary tool is adjusting bank capital requirements for loans involving intangible collateral, potentially forcing banks to hold more capital against such exposures. The Office of the Comptroller of the Currency could also issue guidance limiting national banks' participation in these syndications, a move that would choke off a key source of funding.

Bottom Line

Opaque use in AI lending represents a material and growing systemic risk that exceeds conventional credit concerns.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.