In 2025, artificial intelligence moved from niche tech discussions into everyday life, powering everything from smartphones to enterprise operations. Companies such as AMD, Meta, Tesla, and particularly Nvidia have become central pillars of this AI-driven economy, propelling valuations to unprecedented levels. Yet, as markets celebrate these technological advances, regulators are asking a critical question: who is actually financing the infrastructure behind the AI boom, and what risks might lurk if that debt falters?

Opaque Private Markets Fuel AI Infrastructure

Federal Reserve minutes from January 2026 highlighted a growing concern over AI-related infrastructure financed through private credit markets, which are often opaque and hard to monitor. The Financial Stability Board (FSB) echoed similar worries, citing data gaps in private credit and the potential for systemic risk from concentrated third-party dependencies, cyber risks, and misaligned AI systems. Regulators are increasingly examining these “hidden” channels of financing, as much of AI infrastructure — including data centers and compute clusters — relies on loans and capital commitments that do not trade in public markets.

The Scale and Risk of Private Credit

Private-credit lending to lower-rated borrowers has surged alongside AI-related capital expenditure. S&P Global Ratings notes that US corporate debt rated B‑ and below is set to rise from $56.6 billion in 2026 to $215 billion in 2028, while private-credit loans to this segment already exceed syndicated issuance. Much of this lending is held in closed-end private funds, where valuations are updated infrequently and heavily reliant on internal models, leaving regulators and investors with limited visibility into true exposure.

Unrealized Returns Mask Potential Stress

Studies by Johns Hopkins researchers indicate that a significant portion of private-credit fund performance is tied to unrealized loan values rather than cash distributions. For newer vintages, over 80–90% of reported returns come from residual, still-held positions. Such structures can delay recognition of losses, creating a latent risk if AI infrastructure borrowers struggle to meet obligations. Analysts have also noted that nearly half of direct-lending borrowers report negative free cash flow, further complicating risk assessment.

Limited Secondary Market Raises Liquidity Concerns

Unlike public debt, AI infrastructure loans in private credit have little secondary market, limiting the ability to sell or hedge positions. Jeff Hooke, a researcher in private credit, warns that if borrowers face distress, typical mechanisms such as loan extensions or payment-in-kind interest may delay the appearance of financial stress, masking potential problems from investors and regulators alike.

Regulators Probe the Source of Capital

Beyond individual loans, authorities are increasingly scrutinizing the ultimate providers of private-credit funds. The US Treasury’s proposed Known Investor Program seeks greater transparency on foreign limited partners in private funds, particularly when capital may finance sensitive technology or infrastructure. By identifying “known” investors, officials aim to ensure oversight of the upstream sources of AI infrastructure financing, addressing a long-standing blind spot in the system.

Unanswered Questions Remain

While there is no evidence yet of an immediate systemic threat, regulators are concerned about liquidity, transparency, and potential stress in the private-credit ecosystem supporting AI. Key questions include how quickly problems would surface, how fund valuations might be affected under strain, and which investors ultimately back AI infrastructure loans. Until initiatives like the Known Investor Program are implemented, much of the financing behind the AI boom remains largely hidden.