FinanceFEATURE

Private Credit's Climb to $2.5 Trillion: Why AI Data Centers Are Turning to Non-Bank Lenders

林紀旭 James LinEditor-in-Chief
Published · Updated
Private credit has grown into a global asset class exceeding $2.5 trillion, driven by banks retreating from riskier lending and investor returns climbing toward 12%. A handful of firms — Apollo, Blackstone, Ares, KKR, and Blue Owl — now dominate the market, and Blue Owl has extended it into AI infrastructure with a $27 billion data center financing for Meta, even as the IMF and the Federal Reserve disagree on how much systemic risk this concentration carries.

How Did Private Credit Grow Into a $2.5 Trillion Market, and Who Leads It?

Global private credit assets under management surpassed $2.5 trillion in 2025. The Bank for International Settlements (BIS) put the figure at more than $2.5 trillion, while the International Monetary Fund (IMF) separately estimated the market has grown roughly fivefold since 2009CITE:E1. Both institutions note that the exact number shifts between $1.5 trillion and $2.5 trillion depending on whether unfunded commitments are counted and whether the scope is U.S.-only or globalCITE:E1.

A small group of alternative-asset managers accounts for most of that scale. Based on each firm's own year-end 2025 disclosures, Apollo reported roughly $749 billion in credit assets under management, Blackstone's credit-and-insurance segment (BXCI) reported about $443 billion, Ares reported approximately $407 billion, KKR reported about $322 billion, and Blue Owl reported about $158 billion across its credit platformCITE:E2.

FirmCredit AUM (year-end 2025, self-reported)
Apollo$749 billion
Blackstone (BXCI)$443 billion
Ares$407 billion
KKR$322 billion
Blue Owl$158 billion

The five figures are not perfectly comparable, since some firms' "credit" segments include insurance-related assetsCITE:E2.

How Has Bank Retreat Fueled the Rise of Private Credit?

A measurable retreat by traditional banks from riskier lending has directly fueled private credit's expansion. The BIS found that a one-standard-deviation decline in banking-sector efficiency corresponds to roughly a 33% increase in private credit activity — the single largest factor identified in its analysisCITE:E3. The BIS attributes this to stricter bank regulation imposed after the financial crisis, which pushed financing for higher-risk and mid-sized borrowers toward non-bank lendersCITE:E3.

How Have Rising Returns Drawn Capital Into Private Credit?

Rising returns have pulled investor capital into private credit at an accelerating pace. Preqin estimates that the average internal rate of return (IRR) for private debt rose from 8.1% over 2017–2023 to a projected 12.0% over 2023–2029CITE:E4. Direct lending is predominantly structured at floating rates, and its yields typically run above those of broadly syndicated loans and public high-yield bondsCITE:E4.

How Is AI Data Center Financing Driving Private Credit Demand?

Meta and Blue Owl completed the largest private credit transaction on record to fund an AI data center. In October 2025, the two parties agreed to roughly $27 billion in debt financing for the Hyperion data center in Louisiana, bringing the combined debt-and-equity package to about $30 billionCITE:E5. Morgan Stanley separately estimates that AI data centers, power infrastructure, and fiber networks will require approximately $800 billion in private credit funding between 2025 and 2028CITE:E5.

What Systemic Risks Does Private Credit Pose, and How Do Regulators Assess Them?

The IMF has warned that private credit's opacity and layered leverage could evolve into a systemic risk if left unchecked. It points to relatively fragile borrowers, a rising share of semi-liquid investment vehicles, multiple layers of leverage, valuations that are often stale or subjective, and opaque interconnections among participants — while explicitly stating that these vulnerabilities have not yet become systemicCITE:E6.

The Federal Reserve takes a more measured view. Its April 2025 Financial Stability Report rated corporate credit vulnerability as "moderate" and characterized the associated risk as "limited and contained"CITE:E7. The Fed's assessment notes structural differences from the 2008 crisis: private credit capital is typically locked up for defined periods, leverage and maturity mismatches run lower than at banks, and the sector carries no deposit liabilities subject to runsCITE:E7.

What This Means

The same forces that built a $2.5 trillion market — banks stepping back under tighter regulationCITE:E3 and investors chasing returns that climbed toward 12%CITE:E4 — are now the forces financing individual AI infrastructure deals at a scale, $27 billion in one transaction, that rivals what many of these firms manage in total credit assetsCITE:E5CITE:E2. That concentration is exactly what the IMF's warning is aimed at, even as the Fed currently rates the associated vulnerability as moderate rather than systemicCITE:E6CITE:E7.

📊 Evidence

FAQ

How Did Private Credit Grow Into a $2.5 Trillion Market, and Who Leads It?

Global private credit assets under management surpassed $2.5 trillion in 2025. The Bank for International Settlements (BIS) put the figure at more than $2.

How Has Bank Retreat Fueled the Rise of Private Credit?

A measurable retreat by traditional banks from riskier lending has directly fueled private credit's expansion.

How Have Rising Returns Drawn Capital Into Private Credit?

Rising returns have pulled investor capital into private credit at an accelerating pace.

How Is AI Data Center Financing Driving Private Credit Demand?

Meta and Blue Owl completed the largest private credit transaction on record to fund an AI data center.

📎 Sources

  1. bis.org
  2. sec.gov
  3. preqin.com
  4. cnbc.com
  5. imf.org
  6. federalreserve.gov

Related data

Author's Take林紀旭 James Lin

The Meta-Blue Owl transaction is the clearest sign yet that private credit's growth has moved from corporate lending into infrastructure finance: at roughly $27 billion, a single deal equals more than one-sixth of Blue Owl's entire $158 billion credit platform. That kind of concentration is precisely what the IMF's warning is aimed at — a handful of firms, Apollo, Blackstone, Ares, KKR, and Blue Owl, now underwriting AI infrastructure at a scale where any one deal's leverage or valuation assumptions carry outsized weight. The real question isn't whether this keeps growing — Morgan Stanley's $800 billion estimate for 2025–2028 suggests it will — but whether deal sizes keep scaling faster than the valuation transparency the IMF flagged. Watch whether the next AI data center financing approaches or exceeds the $27 billion Hyperion benchmark, and whether the Fed's 'moderate' rating still holds when it does.

林紀旭 James LinEditor-in-Chief

Related

FEATURE

AI Inference Prices Are Falling About 10x a Year — What the Token-Cost Data Actually Shows

Equivalent-capability AI inference prices have fallen roughly 10x per year across multiple benchmarks, with OpenAI's GPT-4o mini priced about 99% below its 2022 predecessor and DeepSeek-R1 undercutting OpenAI's o1 by roughly 27x. Yet Google now processes about 50 times more tokens monthly than a year earlier, and Gartner projects 2025 global generative-AI spending at $644 billion, up 76.4% — proof that cheaper tokens are fueling more total usage and spend, not less.

Nathan ·
FEATURE

China's AI Playbook Under Export Controls: Ahead on Models, a Generation Behind on Chips

China's AI industry shows a split scorecard: on models, DeepSeek-V3's open-weight, MIT-licensed mixture-of-experts architecture and DeepSeek-R1's benchmark parity with OpenAI's o1 preceded a $593 billion single-day NVIDIA sell-off, while the widely cited $5.576 million cost figure covers only the final training run. On hardware, Huawei's Ascend 910C still delivers about 60% of NVIDIA H100's per-chip performance, and HBM supply constraints cap actual 2025 shipments far below estimated capacity, even as US export controls kept tightening through January 2026.

Nathan ·
FEATURE

Why Are Tech Giants Racing to Buy Nuclear Power? Can SMRs Solve AI Data Centers' Electricity Crisis?

Facing a projected jump in global data center electricity demand from about 415 TWh in 2024 to about 945 TWh in 2030, Microsoft, Google, Amazon, and Meta have all signed nuclear power deals — but most rely on small modular reactors (SMRs) that won't deliver meaningful capacity until 2030 or later, and NuScale's 2023 project cancellation shows the economics are still unproven.

Nathan ·