This. Is. It.

Here is the allocator’s dilemma in two numbers: 10% and 5.8%. The first is what Blackstone was offered to anchor $9 billion of junior debt in Anthropic’s $60 billion chip financing, known internally as Project North Star. The second is what Moody’s now says is the real default rate lurking inside US private credit books, once you stop counting only the losses managers choose to call losses.

Both numbers arrived in the same week. The coincidence is worth sitting with.

What Moody’s Actually Found

Moody’s October 14 report argues that credit stress in the US private credit market may be greater than reported default rates suggest, because reported defaults can understate broader distress among middle-market borrowers when payment-in-kind elections and maturity extensions are excluded from the count. The math is not subtle. The default rate among the 607 private credit borrowers Moody’s assessed stood at 4% in the 12 months to June; that figure rose to 5.8% when PIK amendments and maturity extensions among distressed issuers were included, because the rating agency treats such tools as distressed exchanges equivalent to defaults.

Nearly two percentage points of that gap is not rounding error in a market commonly described as roughly $1.7 trillion. It represents borrowers who are not paying cash interest, not being marked down, and not appearing in any headline statistic that allocators receive.

Business services, consumer services, and software accounted for roughly half of the distressed borrowers identified in the report. That sector overlap matters: software is a large share of many business development company portfolios, and agentic AI has disrupted the seat-based SaaS model that much of that lending was underwritten against between 2020 and 2025.

The Maturity Wall Makes It Worse

Refinancing pressures could add to the strain: lower-rated issuers account for more than half of debt maturities due in both 2027 and 2028 within Moody’s sample, leaving weaker borrowers reliant on extensions, PIK amendments, costly refinancing, or further support from private equity sponsors. In other words, the mechanism being used today to suppress the reported default rate is the same mechanism that will be tested hardest over the next 18 months.

The Anthropic Lens

The Anthropic deal is not itself the problem. Anthropic offered Blackstone a coupon of 10% on $9 billion of junior debt in the $60 billion chip financing known as Project North Star, as reported by the Financial Times citing people familiar with the negotiations. The reported coupon puts a price on Anthropic’s own credit: the FT reported that Blackstone was offered 10% on $9 billion of junior debt backed by Anthropic’s five-year chip-lease payments.

The deal is a clean credit transaction with identifiable collateral and a named counterparty. What it reveals is the pricing signal the market is placing on AI infrastructure risk in late 2026. That pricing signal is high-yield, not investment-grade, even for a company valued at $965 billion in its May 2026 Series H round.

Now ask what the same lenders are charging, implicitly, for the software and business-services loans sitting two hundred basis points tighter on their books, PIK-ing along, marked near par.

The pricing question extends beyond Anthropic. AI infrastructure borrowers financing chip buildouts at high-yield coupons face a structural mismatch if revenue falls short of debt-service costs.

What Investors Are Missing

Level 3 marking and PIK accrual produce stable reported NAVs as credit quality deteriorates. Apollo’s John Zito put the problem plainly: “I literally think all the marks are wrong.” That view predates Moody’s October report. The report gives it a specific number.

The managers most exposed to this gap are not hard to identify.

Stocks to Watch

Blue Owl (OWL) sits at the center of the valuation debate. OCIC saw redemption requests of 21.9% and OTIC saw about 40%, with $4.2 billion left unfulfilled, and OWL stock fell about 66% from its January 2025 peak. PIK exposure and mark stability are the two live questions in every analyst model.

Ares (ARES) carries the heaviest fee-earning asset concentration in private credit among the major alternative managers, making it the most direct earnings play if write-down cycles accelerate into 2027.

Apollo (APO) is both a lender to AI infrastructure, through its involvement in the Anthropic financing, and a manager of middle-market credit books that Moody’s data covers. The tension between those two franchises is underappreciated.

Blackstone (BX) anchored the Anthropic junior tranche and invested $400 million alongside clients to meet redemption requests at its flagship private credit fund earlier this year. Its credit exposure is lower than peers, which is the differentiator if sector stress widens.

BDCs broadly are where the 5.8% number lands hardest. About $20 billion in withdrawal attempts occurred in Q1 2026, and many large non-traded BDCs hit their quarterly repurchase limits, suggesting the semi-liquid retail structure is under systemic rather than isolated stress. The Moody’s report gives that observation a default rate to attach to it.