Notes From the Quiet Unraveling
Notes From the Quiet Unraveling
Monday, April 27, 2026
Nobody panic. The S&P is at record highs. The Nasdaq is up 15% in April. Intel just delivered the greatest single-quarter earnings resurrection since Lazarus put on a hard hat, and five of the Magnificent Seven report this week. Everything is fine. Go back to your screens.
Just try not to look too hard at what's happening in private credit.
Here is a thing that happened this week, reported quietly, processed quickly, and then buried under the Intel surge and the Warsh confirmation and Apple's leadership shuffle: Blackstone's flagship private credit fund — BCRED, the $80.5 billion retail-facing behemoth that was sold to wealth management clients as the intelligent alternative to bonds — disclosed that its non-performing loan ratio hit a record 2.4% in Q1. Two names drove it. Medallia, the software company Thoma Bravo acquired for $5.1 billion in 2021 and which is now heading toward a total loss. And Affordable Care, a dental management services company sitting on $1.4 billion in private loans it can no longer service.
Blackstone marked Medallia's first-lien debt to 60 cents on the dollar. That's down from 78 cents in Q4 last year. Affordable Care went to 70 cents, from 80. These are not rounding errors. These are write-downs arriving in sequence, quarter after quarter, like slow leaks in a hull that everyone on deck has agreed not to discuss at dinner.
The detail that deserves to be read twice: FS KKR Capital Corp valued the Affordable Care loan at 93 cents on the dollar at year-end. New Mountain marked the same loan at 10 cents. The same loan. A 93-cent gap between two professional asset managers holding the same piece of paper. One of them is catastrophically wrong. The question of which one is less interesting than the question of how many other loans across this $1.7 trillion market are sitting on similarly divergent marks that nobody has had to reconcile yet — because the music, so far, has kept playing.
KKR's related private credit fund already had a 5.5% default rate last December. Moody's downgraded it to junk in March. In January, BCRED faced $1.7 billion in net withdrawals — the largest since the fund launched in 2021. Blackstone responded by having its own employees and the firm inject $400 million to staunch the bleeding. Blue Owl Capital ended regular quarterly liquidity payments on one of its retail-facing private credit funds. And yet, last week, Blackstone issued an $850 million five-year bond through BCRED, rated BBB- — the floor of investment grade — yielding 6.2%. Goldman Sachs had done something similar the week before: $750 million, oversubscribed.
The demand was real. That's the unsettling part. Fixed-income buyers, staring at a 10-year Treasury at 4.31% and a Fed about to stop telegraphing its next move, reached for 6.2% from a fund with a record non-performing ratio and a redemption queue. The spread made sense. The underlying made less.
This is what late-cycle credit looks like from the inside: not collapse, but negotiation. Restructuring conversations. Mark-downs in sequence. Liquidity gates framed as investor protection. Bonds issued to reassure the same market you're quietly unwinding into. It is a deeply practiced art, and the practitioners are very good at it, and it tends to work right up until the moment it doesn't.
Meanwhile, Tim Cook is leaving Apple.
Announced last week, effective September 1: John Ternus — hardware boss, mechanical engineer by training, twenty-five years in Cupertino — becomes the eighth CEO in Apple's history. Cook moves to executive chairman. The company published a photo of both men walking side by side at the campus, wearing matching dark shirts, Apple Watches on both wrists, smiling at each other like a succession plan designed by a PR firm that specializes in non-events.
The market took Apple down 2.5% on the news. That is the honest reaction. Cook built a $4 trillion company out of a $300 billion one. He did it by being the world's best operations executive at a time when supply chain mastery and margin discipline were the dominant value drivers in consumer tech. That era is over. AI doesn't care about your logistics network.
Ternus is a hardware man taking the helm at precisely the moment when Apple's hardware advantage — the custom silicon, the tight integration, the annual iPhone cycle — is becoming table stakes. The question investors are actually asking is whether Siri, having embarrassed itself for two years running, can be rebuilt fast enough to matter before the iPhone becomes a beautiful, overpriced terminal for someone else's AI.
2026 is doing something strange to corporate America's corner offices. Greg Abel replaced Warren Buffett at Berkshire. Josh D'Amaro replaced Bob Iger at Disney. John Furner replaced Doug McMillon at Walmart. Now Ternus for Cook. A generation of iconic CEOs is simultaneously clearing out, and the successors share a common problem: they are inheriting companies that were optimized for a world that AI is actively restructuring. The playbook that made their predecessors legendary is not the playbook that survives the next decade.
IBM lost 8.3% last week despite an earnings beat, because its software segment's growth was slowing and investors have decided they'd rather price in its disruption than its resilience. ServiceNow fell nearly 18% after reporting. Not on bad numbers — on forward guidance that wasn't convincing enough that AI wasn't hollowing out its renewal base. These are not small companies. IBM has been declaring itself reinvented approximately once per decade since the 1980s. ServiceNow is a $200 billion enterprise software platform. When the market moves them like that on earnings week, it is passing a judgment.
The judgment is that the value in enterprise software is migrating fast — from the workflow layer to the intelligence layer — and the companies that own the intelligence layer are not IBM and ServiceNow. They are the hyperscalers. Amazon, Microsoft, Alphabet, Meta. They report this week. The consensus capex estimates for those five companies through 2030 have risen 25% since October. The market is voting with extraordinary force that the AI buildout is real, accelerating, and concentrated.
That concentration is worth sitting with. The S&P's April rally is not broad. It is a handful of hardware enablers — Intel, Texas Instruments — and five cloud giants who are simultaneously the infrastructure layer, the platform layer, and increasingly the application layer of the AI economy. Everything else is either a beneficiary or a casualty.
Private credit is lending to the casualties. That's always been the job — take the risk the public markets won't, get paid for illiquidity, hold to maturity. The model works when defaults stay contained. Medallia and Affordable Care are two data points. Healthcare, S&P Global projects, will see credit ratings fall to B and CCC grades across the sector by 2027. Software companies acquired in the 2020-2022 vintage at 15x revenue multiples and financed with private loans at SOFR plus 600 are not, it turns out, the durable cash machines the models assumed.
The retail investor who bought BCRED for the 9.8% yield since inception is still technically ahead. The 7.9% redemption queue — $3.8 billion trying to get out — suggests they have noticed something. Not a crisis yet. A negotiation. A slow, careful, very professional repricing of what illiquid credit in a higher-for-longer world is actually worth.
Watch the marks next quarter. Not the headlines — the marks. The 10-cent and 93-cent valuations on the same Affordable Care loan cannot both survive an audit. When one of them blinks, the rest of the market will have to decide how many of its own loans are priced by faith rather than evidence.
The S&P is at record highs. Everything is fine. The music is good.
Just watch the marks.