Bloomberg Podcasts Sep 20, 2026
With Howard Marks, Co-Chairman of Oaktree Capital Management
Howard Marks says the market has been run by optimists since late 2022, and that the S&P 500 has more than doubled while they have been in charge.
Most people describing a doubled index and an AI boom in the same breath reach for the word bubble. Marks refuses it. He accepts the exuberance and says nobody can yet prove it is irrational, because nobody can say what AI will earn.
"We certainly have exuberance. The question is whether it's irrational."
Howard Marks, Co-Chairman of Oaktree Capital Management, on Bloomberg Podcasts, has spent decades writing the memos that put "irrational exuberance" and the too-hard pile into general investing language, and he was on set at the Qatar Economic Forum having just written two of them on AI.
The full segment is covered here so you can skip it.
Here are the 8 arguments that matter.
Key Takeaways
A central bank that says less makes everyone else safer, because it forces portfolios and businesses to be built for outcomes nobody promised
The market has been ruled by optimists since late 2022, and the S&P 500 has more than doubled in that time
When optimists rule, prices sit high against intrinsic value, which he says calls for "a little bit of caution"
No really bad time since March 2009 is itself a reason for the optimism, because people extrapolate what they have lived through
On AI as a bubble, he will not move from exuberance to irrational exuberance — not because he is bullish, but because nobody can quantify the earnings
He wants a government task force on what AI does to employment, and one question on it: if AI replaces workers, does it replace taxpayers
Oaktree uses AI to marshal data and do early analysis, and has handed it no investment decisions
Stop doing your own critical thinking, he says, and "you'll lose the muscle"
A pure AI company may belong on Warren Buffett's too-hard pile — too hard even for a value investor, and he counts himself one
1. Less Fed Talk Is Better
The segment opened on the Fed. Asked whether he is satisfied with how much the Fed chair is telling markets after the last decision and in the handful of speeches he has given, Marks said he wants less of it.
He is not looking for instructions from the Fed
Well, I actually think that, less communication is better. I'm not looking for the Fed chair to tell me what to do or even tell me what he's gonna do.
Howard Marks
His argument is not about the Fed's independence. It is about what certainty does to everyone else's balance sheet: an investor who believes he knows the path stops building for any other one.
Silence forces people to build for outcomes they were not promised
But I think that if to the extent that the central banker tells people less about what he's going to do, they have to make their portfolios and their businesses, antifragile, which is better for society.
Howard Marks
A society that prepares for one outcome is exposed to every other one
When everybody in society thinks they know exactly what the Fed's gonna do and prepares for that, then the society is dependent on that outcome. And if something else happens, then bad things can happen. So I think it adds to safety in the environment overall.
Howard Marks
2. Let Markets Set Rates
Pressed on whether Kevin Warsh is onto something in letting markets decide rather than leaning on forward projections, Marks said his preference is for a central bank that intervenes rarely.
Leave the economy alone unless it is running hot or cold
Well, my personal preference is for a less activist central bank that normally lets the economy do its thing unless it's in danger of going off the rails too hot or too cold. Yeah. If it's not too hot or too cold, leave it alone. So, I'm in alignment on that.
Howard Marks
He then defined the rate he would rather see set: the one that comes out of borrowers and lenders bargaining with each other, with no policy rate standing behind it.
The natural rate is what a negotiation produces
We talk about something called natural interest rates. Those are the rates that result from negotiations between borrowers and lenders. I think that's a great idea. This is called the free market. Let the market do its thing.
Howard Marks
He finished the point on himself, and on his own mortgage.
He would negotiate his own rate down
Listen. I'm a tough negotiator. Anything that can get that mortgage rate down, you and I will go at it.
Howard Marks
3. The Optimists Are In Control
Asked what this environment is like for a value investor, Marks first corrected the label. He uses a lower-case v: the job is estimating what an asset is worth and buying below it, not screening for cheap statistics.
Value investing, small v
Well, look. I am what is called the value investor, not with a capital v, but a small v.
Howard Marks
The whole method is one sentence
And that means our approach is to figure out the intrinsic value of an asset and see if we can buy it at a fair price relative to that value. That's all it means, in my opinion.
Howard Marks
Then the market call. He dates the current regime to late 2022 and says the index has more than doubled since.
Optimists have been running the market for three years
And the challenge today is that in my opinion, since roughly 10/01/2022, I believe that the markets have been generally ruled by optimists. And the S and P has more than doubled in that period. It's been a very strong period.
Howard Marks
The consequence, in his framework, is mechanical rather than moral: optimism lifts price relative to value, and a higher price against the same value is a lower margin of safety.
High prices against value are a reason for caution, not alarm
And I think that, you have to recognize that when the optimists rule, prices tend to be high relative to intrinsic value, which means that it's a time for a little bit of caution. It's not the worst thing I've ever seen, but I do think it's important to note that the optimists have been in control.
Howard Marks
4. Why Nobody Is Pessimistic
The anchor, who covers diplomacy, put the puzzle from her own beat: markets rally on any hint of a deal out of Washington, Europe or the Middle East, and the people who cover those talks know most of them will not land. Why does the optimism survive?
Marks gave four reasons, and started with the ones he thinks are real.
The US economy is the best of the developed world, and there is no alternative
Well, that's a great question. But I think there are certain underlying truths which are positive, which are in the ascendancy. Our economy is doing pretty well in the aggregates. Yeah. It is, in my opinion, the best performing developed world economy. So? So, the old Tina trade, there is no alternative. And there is the excitement over AI.
Howard Marks
The fourth reason is the length of the good run itself.
Nothing really bad has happened since March 2009
Oh, and the other thing is that we haven't really had a really bad time since March of o nine when the stock market bottomed during the global financial crisis. And so people get used to good times and they extrapolate good times, and that contributes to the optimism.
Howard Marks
The anchor put the behavior to him as asymmetry: optimists discount the bad news. He agreed and sharpened it.
Good news gets priced, bad news does not
Yeah. And they get excited when there's word of a settlement. They don't get depressed when it falls through.
Howard Marks
What a positive bias is worth doing about
So you have to look at that, and you have to say, well, there's a positive bias in the market today. And when there's a positive bias, you should acknowledge it. And what do you do about it? Maybe you build in a little defense, or may you take care in what you do.
Howard Marks
5. Nobody Is Working On Jobs
The anchor turned to a memo two back, in which Marks wrote about what AI would do to people outside the big cities if the technology delivers what has been promised, and asked whether the fears he set out then have got worse.
His answer separated the thinking from the worrying, using his wife's line for the difference.
Worrying is bleeding before you're shot
Well, I think that a sentient human being has to think about these things, has to acknowledge there's nothing we can do about it. And when I reach that point in my own process, I kinda let it go because my wife defines worrying as bleeding before you're shot.
Howard Marks
Asked whether enough people are thinking about it in the right way, he said the public is; the government is not.
There is no government task force on AI and jobs, and there should be
Well, people are. On the other hand, I don't see anything out of government. I think there should be a task force on the effect of, the coming effect of AI, and there isn't.
Howard Marks
He then gave the task force its first assignment, which is a fiscal question rather than a labor one.
If AI replaces the worker, does it replace the taxpayer
If people are put out of work, they won't pay taxes. If AI takes their place as workers, will it take their place as taxpayers and how? So what will happen to government revenues when and if people lose their jobs?
Howard Marks
6. Oaktree Uses It For Data
Asked whether he uses AI himself and whether Oaktree has put it into the business, Marks drew the line at the point where judgment starts.
It does the data gathering and the first pass, not the decisions
Oh, well, Oaktree certainly has integrated into the business, and we use it to marshal data and do early analysis and it was a task that used to take a long time. We certainly haven't turned business decisions over to it, investment decisions.
Howard Marks
His reason is a claim about where an investor's advantage actually comes from. Not the data, and not the speed of processing it, but the judgment of how much a given piece of data matters.
The advantage is in weighing the data, and he says AI cannot do that yet
And I think that to the extent you wanna be an exceptional investor, the exceptionalness doesn't come from some computer or algorithm. The exceptionalist comes from being able to do a superior job of, assessing the import of the data. And that superiority, I don't think, is available through AI yet.
Howard Marks
Asked whether younger hires are handing their critical thinking to the machine, he said Oaktree discourages it, on the grounds that the skill decays if it is not used.
Stop thinking and you lose the muscle
We don't encourage that. Okay. There's a lot of do, so I was wondering. Yeah. There's a lot of work to be done. But number one, if you stop doing critical thinking, you'll lose the muscle, I think.
Howard Marks
His analogy for what AI does to the active management industry is index funds, which took the business from most active managers without taking it from the best.
Like indexation, it clears out the middle and leaves the top
And I think that kinda like indexation did for equity investing, it'll put a lot of active investors out of business, but not the best.
Howard Marks
7. Exuberance Without Proof
The anchor said it would be useful to spot a bubble in real time, which Marks has written cannot be done, and asked how bubbly this one looks given the talk about circular financing among the AI companies.
Marks went back to the phrase Alan Greenspan used in 1997 and took it apart into its two words.
Exuberance is established; irrational is not
Well, look. The main thing is, as I wrote in the last memo on the subject, Allen Greenspan used the phrase in '97, I think it was irrational exuberance, and that is the definition of a bubble. We certainly have exuberance. The question is whether it's irrational.
Howard Marks
The reason he will not supply the second word is that the case for irrationality requires knowing what AI will earn, and nobody does.
Nobody can specify what AI will do, so nobody can call the price wrong
And since nobody, in my opinion, can spec out exactly what it is that AI has been gonna be doing and with what effect and with what profitability, I don't think anybody can say, that it's irrational, but you have to consider the possibility.
Howard Marks
What that leaves an investor is a question about appetite rather than about valuation.
The real question is how much of the aggressive investors' money you need to make
And so, when you tackle AI from an investment point of view, you have to think about how confident you are that in your judgments and how much like in everything else but taken to a higher power, we have to think about how much how important is us to make all the money that's gonna be made by the most aggressive people, and are we willing to take on the uncertainty which is associated with that?
Howard Marks
8. AI May Be Too Hard
Asked whether a value investor can warm to AI companies at all, Marks said it is challenging and then split the field in two. The hyperscalers, he said, have other businesses that produce an earnings number to work from.
Microsoft and Amazon have businesses you can put a number on
Microsoft and Amazon and the like, They have other businesses which will support, for example, an earnings estimate.
Howard Marks
A company whose entire model is AI gives a value investor nothing to anchor to, which puts it where Warren Buffett said such things go.
A pure AI company belongs on the too-hard pile
But if you're looking at somebody who's just in the business of AI, and if they're all in on AI, their entire business model is dependent on AI. Maybe they haven't produced a profit yet or have, I just think that there's an unusual degree of uncertainty in coming up with any quantification of earnings potential and thus of what we call intrinsic value. I think it's gonna be hard. And, Warren Buffett always said, oh, we put that on the too hard pile. It may be too hard for the value investor, certainly the one with the capital v.
Howard Marks
The anchor pushed: if all of that is unknowable, is the exuberance not irrational by definition? Marks said no, twice, and the exchange got short.
Not knowing something is not evidence of anything
Don't know it doesn't mean it's that something's true.
Howard Marks
He then laid out the three cases as a ladder of certainty: a traditional company whose earnings can be estimated, a tech company where the same estimate is less reliable, and a pure AI company that has never reported results.
Nobody can say what a company with no results earns in five years
But, a pure AI company, which has never announced results, you and what are they gonna make five years from now? You have to say, well, there's an unusual amount of uncertainty.
Howard Marks
The conclusion is the opposite of paralysis. Uncertainty changes the form and the size of the position, not whether there is one.
Uncertainty sets the size, not the decision
The fact that there's uncertainty present doesn't mean you don't do something. But you acknowledge it and that may inform how you do it, which form of AI investing you do, and it may inform how much you bet on that.
Howard Marks
And it is not a reason to sit out
But the fact that there's uncertainty involved cannot keep you from doing something.
Howard Marks
Bonus Insights
He says there is no rampant pessimism anywhere today
Marks listed the absence of pessimism alongside the four positive drivers as part of why prices sit where they do, rather than as a separate warning.
He accepts there is nothing he can personally do about AI's effect on work
And in this case, there's nothing we can do about it. Nothing certainly nothing I can do about it personally. So, you roll with it. That's all. You acknowledge it.
Howard Marks
Marks's bottom line is that the AI trade is not a bubble he can prove and not a price he can check: the exuberance is real, the earnings are unknowable, and what an investor does about that is decide how much of the aggressive money he needs to make and size the position accordingly, while the broader market carries a positive bias that argues for a little defense.
Products, Companies & Tools Mentioned
Oaktree Capital Management (His firm, which has put AI into data gathering and early analysis and has handed it no investment or business decisions)
Microsoft and Amazon (His examples of hyperscalers, which he says have other businesses that support an earnings estimate — unlike a company that is all in on AI)
ChatGPT (The anchor's shorthand when she asked whether he uses AI in daily life)
The S&P 500 (His measure of the optimists' run: more than doubled since roughly late 2022)
Books & Resources Mentioned
Howard Marks's Oaktree memos (He refers to the last memo on AI and to one two before it on what AI would do to employment outside the big cities)
Listen to the full episode
🔴 YouTube
Watch the full episode:
More Howard Marks interviews and podcast appearances
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: