The AI Bubble Is Bigger Than You Think

Private credit—now a $1.6 trillion industry—has become the primary financier of the AI infrastructure boom through SPVs and securitizations that push debt off Big Tech balance sheets (eg, Blue Owl’s majority stake in the $30 billion SPV for Meta’s Hyperion), enabling a rebranded, lightly regulated “shadow banking” model (Blue Owl itself manages about $295 billion and has recently blocked redemptions after a fund merger). The deals rest on fragile economics: GPUs may depreciate in ~2 years yet are being financed on much longer schedules and securitized (CPE estimates 61% of relevant securitizations are data-center related), companies are extending depreciation and leveraging hardware to buy more hardware, and even major AI players show huge cash burn (OpenAI lost $11.5bn last quarter), creating a high risk of stranded assets and overstated cash flows. Given banks’ links to private credit (Moody’s cites at least $300bn in bank loans tied to private credit), rewritten covenants, and wide broker-dealer participation, the sector poses systemic contagion risks to institutional portfolios and retirement plans if the data-center/private-credit bubble bursts.

Analysis

Private credit has become the principal financier of the AI infrastructure build-out, with industry assets under management cited at $1.6 trillion (February) and major players like Blue Owl managing roughly $295 billion and holding the majority stake in Meta’s $30 billion Hyperion SPV. The article documents operational strains — Blue Owl’s post-merger redemption block that can impose ~20% losses on investors and a reported 61% share of relevant securitizations tied to data centers — that signal illiquidity and concentrated risk in these SPV-backed structures.

Economic mechanics are fragile: GPUs that may sustain heavy AI training only ~two years are being financed on extended depreciation schedules, firms are leveraging GPUs to buy more GPUs, and OpenAI’s $11.5 billion quarterly loss highlights extreme cash burn versus future revenue visibility. These mismatches create a credible pathway to stranded assets and overstated cash flows as data centers and bespoke power plants age or become obsolete.

Systemic contagion risk is nontrivial because banks carry at least $300 billion in loans to private credit (Moody’s), covenants are being rewritten to shield private-credit sponsors, and broker-dealer distribution links place retail and 401(k) investors at potential risk; Blue Owl’s stock weakness (down sharply year-to-date and off 6% in a single session) and headline warnings from industry executives amplify downside signaling.

Key Decisions for Investors

• Reduce or avoid incremental allocations to private-credit funds and SPVs that are explicitly tied to data-center or GPU-backed securitizations given demonstrated illiquidity and redemption risk
• Underweight or hedge exposure to Blue Owl (OWL) and other firms with large private-credit footprints while reviewing fund terms for gating/redemption clauses,
• Stress-test portfolios for second‑order exposure via banks and 401(k) vehicles (Moody’s cites ~$300bn bank links) and increase liquidity buffers or buy protection where available,
• Monitor short‑term indicators — redemption blocks, covenant amendments, securitization issuance tied to data centers, GPU depreciation assumptions, and large cash‑burn disclosures (e.g., OpenAI) — and be prepared to de-risk on deteriorating readings