Oaktree's Panossian: This Is a Correction, Not a Crisis

What's New

In a candid Bloomberg TV appearance, Oaktree Capital co CEO Armen Panossian delivers what may be the sharpest framing yet of the private credit selloff: this is not a private credit problem, it is a vintage problem. With Blue Owl reporting 41% redemption requests on one of its BDCs and publicly traded BDCs trading at steep discounts, Panossian argues the market is painting the entire asset class with one brush when the real issue is concentrated in pre 2022 vintages loaded with software exposure and non cash pay structures. His warning, however, is not purely reassuring. He believes we are in the early innings of the correction, that AI risk in software portfolios is likely bigger than most managers have acknowledged, and that the real threat is not redemptions themselves but what happens when leverage providers to BDCs start pulling back.

Why It Matters

• Vintage explains dispersion: The dividing line is November 2022, when ChatGPT launched. Pre 2022 vintages were built in a world of ultra low rates where software was a darling sector and AI disruption was not part of the underwriting thesis. Post 2022 vintages carry meaningfully different risk profiles. The market is not distinguishing between the two.

• Leverage providers are the real risk: While semi liquid BDC redemptions alone are unlikely to force fire sales (managers can cap at 5%), the bigger concern is banks tightening leverage tolerance on BDC vehicles. If lenders mark software collateral down and cut exposure, managers face margin call dynamics that require fresh equity, deleveraging, or forced asset sales.

• Software exposure is understated: Many portfolios classify software companies under labels like business services or government services. When you adjust for these misclassifications and layer on leverage of 1x or more against portfolios with 20 to 40% software concentration, even a modest 5 to 10% tranche of true AI risk can produce a 10 to 15% hit to equity.

Big Picture Drivers

• Retail capital created deployment pressure: Five to six years of strong inflows, compounded by persistently low M&A and LBO volumes (COVID, rate shock, geopolitical disruption, Liberation Day tariffs), created a mismatch where managers had to put money to work in a deal starved environment. Excessive risk taking was the predictable result.

• PIK and ARR loans are ground zero: Oaktree avoided recurring revenue loans and paid in kind structures for years because the compensation did not match the risk. With hindsight, that discipline proved correct. As claims grew through accrued non cash interest and valuations collapsed, loan to value ratios on these deals have blown out far beyond original entry levels.

• Marks may not be done falling: Panossian is direct that nobody knows how deep the AI impact goes. Agentic AI's ability to potentially replicate legacy SaaS products means some software companies face binary outcomes. It is too early to say we have seen the lowest marks.

• Macro tail risks are building: A prolonged Iran conflict keeping oil meaningfully above $100 could trigger recession. Parts of Asia are already experiencing recessionary conditions from disrupted refined product flows through the Straits of Hormuz, and even a quick resolution carries a lag effect as production restarts.

By The Numbers

• 41%: Redemption requests reported by one Blue Owl BDC

• 5%: Maximum quarterly repurchase offer most semi liquid BDCs can provide

• 20 to 40%: Range of software exposure across direct lending portfolios, with some vehicles at the high end

• 10 to 15%: Potential equity hit from AI related software losses on a levered basis, even if only 5 to 10% of the portfolio faces true disruption

• $1.8T: Total private credit market size, of which semi liquid BDCs and their software loans represent progressively smaller subsets

Memorable Quotes

• "This is a correction. I don't think this is a systemic rejection of the private credit asset class." — Panossian, framing the current environment as a painful but contained recalibration rather than an existential threat.

• "We couldn't predict what happened now, but we did prepare." — Panossian, on Oaktree's decision to maintain conservative leverage, hold liquid assets, and avoid chasing growth, which now positions them to play offense.

• "Nobody knows how bad it can get at this point." — Panossian, on why it is premature to assume current software markdowns fully reflect AI disruption risk, given the pace of agentic AI development.

• "Any software program that you use day to day, you could go into it, bring an AI agent with you, and say can you just recreate this for me. And it's not too far out of the realm of reason to think that could actually happen." — Panossian, illustrating the existential nature of agentic AI for legacy SaaS businesses in the most concrete terms.

• "Anytime that in a short period you do get meaningful inflows and cash drag creates a big problem for returns, then you do see excessive risk taking in any asset class." — Panossian, explaining how the structural incentives of retail evergreen vehicles, where managers must deploy immediately or suffer cash drag, drove the underwriting deterioration now coming home to roost.

The Wrap

Panossian's interview is notable for what it does not do: sugarcoat. Unlike many industry participants offering reassurance through averages and historical context, he names specific mechanisms that could deepen the correction, from leverage provider pullbacks to understated software exposure to the genuinely unknowable trajectory of agentic AI. His central thesis that vintage explains the dispersion is both clarifying and sobering. It means the problem is not private credit as a structure but the specific decisions made during 2019 to 2022 when capital was abundant, software seemed invincible, and nobody was stress testing for AI displacement. For Oaktree, the payoff for years of discipline is an opportunity to deploy into dislocation. For the broader market, the question Panossian leaves unanswered is whether the correction stays orderly or whether a confidence crisis, triggered by geopolitics, leverage provider retrenchment, or simply more bad headlines, tips it into something worse.