Oaktree's Howard Marks on Unpredictability, Importance and Investing in AI
Howard Marks, after nearly five decades lending to below-investment-grade companies, now faces what he calls the most unpredictable environment of his lifetime. The very power that makes artificial intelligence important also makes it impossible to forecast — raising existential questions about employment, corporate survival, and the validity of 30-year bonds issued by tech giants. Meanwhile, a once-hot asset class — private credit — has shifted from «what about private credit?» (enthusiastic) to «what about private credit?» (anxious), as early advantages erode and retail investors discover illiquidity the hard way. Can traditional investment judgment survive when AI can read everything, remember everything, and extrapolate patterns faster than any human — yet lacks the intuition to sense when «the hairs on the back of your neck stand up»?
Pontos-chave
AI is rendering the investment world far less predictable than at any prior point, because the same transformative power that gives it importance also makes its consequences unknowable.
Private credit has moved from lush early-mover advantage (high rates, strong safety) to equilibrium with public credit, as competition bid down spreads to roughly 125 basis points — adequate but no longer special.
Lending to AI or tech companies for fixed returns is likely a mistake; if you're taking fundamental business-model risk in an unpredictable environment, you should own equity and capture upside, not accept bond coupons.
The worst loans are made in the best of times: 17 years of low defaults and easy money have bred complacency, FOMO, and lax underwriting — setting the stage for an eventual default cycle.
AI is a powerful research and pattern-recognition tool, but it makes predictions (hypotheses) rather than judgments; humans still need to assess character, exercise intuition, and decide when assets are cheap enough to buy aggressively.
Em resumo
Marks believes AI's sweeping, unpredictable impact renders most forward-looking predictions unreliable, yet he remains confident that assets can still become cheap enough to warrant aggressive buying — so long as human intuition, not algorithmic prediction, guides the final call.
Private Credit: From FOMO to Fear
Private credit has shifted from lush advantage to equilibrium pricing.
The Most Unpredictable Environment Ever
AI's transformative power makes the future impossible to forecast with confidence.
Marks argues that artificial intelligence has introduced a degree of uncertainty unprecedented in his 48-year career. The very capabilities that make AI important — its ability to displace labor, render software obsolete, and reshape entire industries — also make it impossible to model with any reliability. «We've never contemplated questions like these,» he notes, pointing to the potential for mass unemployment, societal upheaval, and the collapse of business models that seemed durable just months ago.
This unpredictability has profound investment implications. Marks has always maintained that successful investing requires two things: a view of what will happen and an honest assessment of the probability that view is correct. In the age of AI, he believes the second element — confidence in one's forecast — has collapsed. The result is a world where lending money to companies for 30 or 40 years (as Google, Microsoft, and Amazon have done) looks less like prudent stewardship and more like «optimism and credulousness in the ascendancy.»
For Marks, this is not a reason to abandon investing, but it is a reason to demand a higher margin of safety, to favor equity over debt when taking business-model risk, and to wait for assets to become genuinely cheap before deploying capital aggressively.
Lending to AI Companies: A Mistake
If you're taking fundamental business risk, own equity — don't lend.
Lending to AI Companies: A Mistake
Marks quotes Oaktree colleague Bob O'Leary: if you're going to finance AI or tech companies in this environment, «you should probably buy the stock and get the upside if it ensues, rather than just lend the money and get a fixed return if they're successful.» Lending for fixed coupons while shouldering existential business-model risk is a mismatch — you bear downside without participating in upside.
What AI Can and Cannot Do for Investors
The Worst Loans Are Made in the Best of Times
Seventeen years of low defaults bred complacency and set up the next cycle.
Marks observes that the financial markets have enjoyed an extraordinary run since March 2009 — 17 years with only brief interruptions (three bad weeks in March 2020, a poor 2022). During such periods, money flows freely, lenders compete for deals by lowering standards, and FOMO replaces skepticism. «The worst loans are made in the best of times,» Marks repeats, because easy money creates the conditions for undeserving credits to get funded and frauds to be perpetuated.
He points to the bankruptcies of First Brands and Tricolor in late 2024 as early warning signs — «cockroaches in the coal mine.» These failures, some involving probable fraud, are the natural result of a long period of lax underwriting. Marks expects more shoes to drop: «It's only when the tide goes out that we find out who's swimming naked,» borrowing Buffett's phrase. Yield spreads on sub-investment-grade credit remain at the low end of historical ranges, suggesting the market has not yet priced in elevated defaults.
Key Figures from the Conversation
Notable data points illustrating market conditions and AI's impact.
Waiting for the Buying Point
Marks remains cautious until assets become cheap enough to warrant aggression.
“I'm always cautious until the time when I believe... that caution has to be overcome. And by the way, a lender should be 80% caution. So I think we'll remain cautious until we believe that the disappointment and the declines have been profound enough to make it time to be aggressive. And then hopefully, as we have in past cycles, hopefully we'll be the best, most aggressive people on the planet. And, by the way, may be the only aggressive people left on the planet.”
Pessoas
Glossário
Aviso: Este é um resumo gerado por IA de um vídeo do YouTube para fins educacionais e de referência. Não constitui aconselhamento de investimento, financeiro ou jurídico. Verifique sempre as informações com as fontes originais antes de tomar decisões. O TubeReads não é afiliado ao criador do conteúdo.