Will Blue Owl's software credit nonaccrual rate remain below 1% through FY2026?

The 30% average LTV provides a 70% equity cushion — for a loan to become nonperforming, the borrower's enterprise value must decline by 70%, which is extremely rare for companies with ~$300M EBITDA growing mid-teens. The 8 bps historical loss rate and current 0.1% nonaccrual rate would need to increase 10x to reach 1%. Even if AI disruption accelerates, the 3-year average remaining duration means most loans mature before any disruption thesis fully plays out. The base rate for alternative lending nonaccrual spikes of this magnitude is very low.

While the mathematical protections are strong, the Black Swan Beacon correctly identifies that 6 of 7 lenses share the assumption of historical-range performance. AI disruption is a genuine structural force — not a cyclical one — and historical loss distributions may not apply. The 8% AUM concentration in one sector facing technological upheaval creates correlation risk that independent loss models would miss. However, 1% nonaccruals is still a very high threshold given starting conditions. Even a material deterioration from 0.1% to 0.5% would still resolve YES.

The compound scenario — software defaults triggering BDC redemptions triggering forced sales triggering more defaults — is theoretically possible but requires multiple low-probability events to cascade within the FY2026 timeframe. The 60% origination from existing relationships and management's incumbency advantage suggest they have good visibility into borrower health. Management's emphatically defensive posture is a weak negative signal but does not overcome the strong quantitative protections. The more likely risk is narrative-driven stock pressure, not actual credit losses.

Current nonaccruals at 0.1% means they need to increase 10x to breach the 1% threshold. With 30% LTV and $300M average borrower EBITDA, these are not marginal companies — they are substantial software businesses with real cash flows. AI disruption fears are valid at a 5-10 year horizon but unlikely to cause nonaccrual spikes within the next 4 quarters. The mid-teens EBITDA growth in Q4 2025 shows these borrowers are currently thriving, not declining.

Software credit has performed well through multiple cycles, and the LTV protection is genuine. However, I weight the concentration risk somewhat more heavily — 8% of AUM in one sector is significant. The management defensive tone ('no red flags, no yellow flags, largely green flags') with extended monologue defending the portfolio suggests they are actively managing this perception. While the math favors staying below 1%, a single high-profile software company restructuring could change the narrative dramatically even if it does not breach the threshold.

The question is specifically about nonaccruals breaching 1%, not about whether any deterioration occurs. Given the 30% LTV, a borrower would need its value to decline substantially before Blue Owl's loan becomes impaired. Even in the worst-case AI disruption scenario, the degradation would be gradual — SaaS companies retain recurring revenue even as growth slows. Mass nonaccruals require mass enterprise value destruction, which typically takes years, not quarters.

0.1% current nonaccruals, 30% LTV, 70% equity cushion, mid-teens borrower growth. Getting to 1% requires catastrophic sector-wide failure in 4 quarters. Extremely unlikely given current metrics.

Strong quantitative protections make 1% breach unlikely. AI disruption is a real long-term risk but would need to materially impair borrowers with $300M+ EBITDA within FY2026. 3-year average duration means most loans won't mature during the risk window. Concentration risk is the main concern but not sufficient to move probability below 80%.

The math strongly favors remaining below 1%. Even accounting for tail risk scenarios (compound cascade, clustered defaults), the threshold is high enough that only a catastrophic, sector-wide event would breach it. Base rate for this magnitude of deterioration in one year is very low.