The Shiller CAPE ratio on the S&P 500 is above 42. Jeffrey Gundlach says that every time it has been above 35, the next ten years of real returns have been negative, with no exceptions.
Most managers respond to that by underweighting expensive stocks. Gundlach has done something narrower and more specific: he has rebuilt his recommended portfolio so that it holds almost no artificial-intelligence exposure at all, and he did the last of it in the week before this interview.
"So everything that I'm recommending is completely separate from AI exposure."
Gundlach has run money for more than forty years and called the secular bottom in interest rates in 2020; DoubleLine, which he founded, publishes the allocations he describes here every quarter.
The full interview is covered here so you can skip it. 62 minutes of audio, 22 minutes of reading.
Here are the 14 calls that matter.
👤 Guest: Jeffrey Gundlach, founder and CEO of DoubleLine Capital, who called the secular bottom in interest rates in 2020 and publishes a quarterly allocation webcast for the firm's investors
🎙️ Host: Julia La Roche, a financial journalist who runs a daily interview show on markets and the economy
📰 Published: 16 September 2026 on YouTube · recorded before the Fed's September decision
🔴 YouTube | ⏱️ 1 hr 2 min | ✅ Time saved: 40 min
Key Takeaways
A Shiller CAPE above 35 has never been followed by a positive real 10-year return, and it is above 42 now
At 2% inflation that implies negative nominal returns for a decade
His recommended portfolio now holds essentially no AI exposure, including in the equity sleeve
The 30% equity allocation is a single equal-weighted index of the largest companies by revenue
Credit is cracking along AI lines, not along ratings lines
AI-linked junk bonds are 50bps wider than their tights and AI bank loans about 130bps, while the rest of high yield has barely moved
He thinks ratings are being bought rather than earned, and that the insurance industry is where it detonates
One firm's affiliated insurers went from 3% affiliated investments to about 42%
He expects the next CPI print to start with a 4 and stay above 4 through March
30-year TIPS do not protect an investor from rising rates, and have not for six years
Nominal and TIPS yields both rose 500bps; the breakeven between them barely moved
He wants a 2% real yield before extending maturity, which means buying the long bond at 6%
$6B of Treasury buybacks is one day of the deficit, so the only real options are inflating the debt away or restructuring it
1. CAPE at 42
La Roche opened on the big picture, and Gundlach started with valuation.
The Shiller cyclically adjusted price-to-earnings ratio on the S&P 500 is above 42. His claim about what that has meant historically is unqualified: "And any time that it has been 35 or higher, every single time, the forward 10-year return in real terms, so versus in inflation adjusted has been negative."
The most common outcome he cites is about −5% real per year. If inflation runs at 2%, as Kevin Warsh says it will, that means negative returns in nominal terms for a decade — and, he said, "there are no exceptions to it."
Valuations have kept rising while the competing yield rose with them. Treasury yields are up about 75 basis points since his previous appearance six months earlier, and the stock market is higher anyway.
2. The Mood Turns
Gundlach's framing for the moment is social rather than financial: a story that everyone believed is now being told the other way round.
The AI narrative flipped inside a week, from a technology that would make everyone rich to one that would kill everybody by the end of the decade. He called it strange behavior — something rolls along, a great deal of money is made in it, and then the narrative changes radically.
He drew the parallel to private credit: "It's kind of like how it was when a year ago everything kind of changed regarding the perceived success of private credit."
"When you know the numbers have been reported and looking back now it's fairly clear that the numbers that were reported a year ago were not accurate which means that the performance reported for the last few years is not correct."
The host explainer here is Gundlach's own, and it is the best anecdote in the interview. He grew up in Buffalo, New York, where a steel mill had been moved from Scranton, Pennsylvania. "It was over a mile long and everything's great until about 6 years later when the depression of 1907 showed up and nothing had changed at the steel plant except the press went from isn't this wonderful to reports on injuries, deaths, limbs lost, people crushed, burned to death at the steel mill."
"It went from let's all look at the bright side of things to wait a minute, it isn't all perfect." Nothing at the plant had changed. The coverage had.
3. Cracks in the Triple Cs
What he is watching most closely now is the credit stack, sliced finely enough to see where the damage starts.
A few months ago every tier was tight. Now the lowest-rated tier is eroding: "For example, the weakest bond market sector, and you we're slicing them very thin, is triple C bank loans, which are down several percent in price and down in about five or 6% in total return, while higher rated bank loans are still doing fine." Higher-rated loans are up about 4%.
Split high yield into AI-related borrowing and everything else and the difference is visible. "The junk bonds are out about 50 basis points from their tights on the AI and the bank loans are out more like 130 basis points from their tights." Non-AI spreads have barely widened and bank loan prices are near their peak. It has not reached the single-B category yet, which is what he is waiting on.
"We're starting to see that the market doesn't believe the ratings of some of these companies."
His two examples are recent deals. "It's like when SpaceX borrowed a bunch of money, the bonds widened out to levels about three notches lower in credit quality." SpaceX was rated BBB−, the lowest investment-grade rung, and he said he suspects the rating agencies were persuaded into it. An Oracle issue rated as junk widened sharply straight after it was sold.
"What the bond market is saying is these ratings don't make sense to us."
4. Ratings for Sale
From there he moved to how the ratings themselves are produced, and this is where he is most direct.
"And we've seen that in particular in growing strains in the private credit market where it's become people are becoming aware that there are seven or eight rating agencies and they are the private credit firms and the insurance companies that they own, they're able to arbitrage these ratings." A borrower can collect several ratings and use the best one.
His more cynical version, which he said is now worth entertaining: "It might be that you just get a price list." He priced the satire out loud — a C rating costs a dollar, "You want a triple C rating, it's $10," a BBB− costs a million.
The point of buying the rating is capital treatment. A higher rating means less capital held against the asset, which means more leverage at the insurance company — amplified again, he said, by leverage at the reinsurance companies that is not even funded.
5. The Four-Part Portfolio
Gundlach walked through the allocation he publishes in his quarterly webcast, which exists because investors kept asking which DoubleLine funds to buy and he cannot answer that in a regular market call. It is the most concrete part of the interview.
Equities: 30%, down from 40% the first two times he did this, and all of it in one holding — an equal-weighted index of roughly 440 of the largest US companies ranked by revenue rather than earnings. "You have almost nothing. You're being as far away from that as you can."
Fixed income: 30%, split as a barbell. Half sits in his total return fund, which holds no corporate bonds at all, let alone AI bonds. The other 15% goes to the riskiest thing he could pick — local-currency emerging market debt yielding over 7%, where he expects to make money on the currency as well because he thinks the dollar is heading lower.
"And I've only allocated to local currency emerging markets once in my career. That's this time" — he first did it in June of last year, and it has been the best-performing fixed income sector
Real assets: 20%. Half is gold, where he has moved around a lot: 25% a year ago, cut to 5% when gold went above $5,000, and back to 10% now that it is at 4,300. The other half is DoubleLine's rules-based commodity ETF, which rebalances monthly. "That's up 38% year-to date. So that's doing awfully well."
Dry powder: 20%, and not in cash. "But I don't just buy cash. I don't like cash. I think you can get a spread above cash." Half goes to a commercial real estate ETF he described as very carefully managed, high in the capital structure, duration of two and yielding about 6%; half to his flexible fund, which is benchmarked against both cash and the Bloomberg bond index.
The whole package yields about 6.25% with a duration of two, which he measures with the Sherman ratio — yield divided by duration. If yield and duration are equal, a 100 basis point rise wipes out the income. "But if you have a yield of six and a quarter and a duration of two, rates would go up 200 basis points and you'd still be positive."
"I use the analogy of loading the dishwasher after a big dinner party. You have to make it all fit together."
The AI exclusion is recent and deliberate. "And I was perfectly fine owning some AI by using other types of equity vehicles." That changed the week before the interview.
6. $41T and Operation Twist
La Roche turned to rates, noting that Gundlach called the secular bottom in 2020 and that the 10-year has been above 5%.
He had said six months earlier that $40 trillion of federal debt would be a psychological level the way $4 gasoline was. He watched the pace of accumulation, forecast Halloween, revised it to Labor Day, "And amazingly the next day they announced it had gone over 40 trillion." In a meeting on the day of this interview, someone quoted 41 trillion.
The day after that announcement, the Treasury secretary floated Operation Twist — issuing large volumes of bills and buying long-dated Treasuries. "He didn't buy anything. He didn't do it."
Gundlach does not accept the stated rationale. The official explanation is liquidity in off-the-run Treasuries, but the bid-offer difference between an off-the-run and an on-the-run bond is a couple of basis points. "So, I think there's something else going on there. And I think it might be just, a full-on Operation Twist is what they really want to do to control long-term rates."
The buybacks so far are too small to matter. "So six billion isn't going to do anything. That's basically one day of the deficit." The program went from $2 billion to at least $4 billion to $6 billion; the market rallied for about 30 seconds on the announcement and sold off again.
"I've said in the past that rates will naturally go up if they're left to market forces, and that certainly has happened." With the deficit heading toward $50 trillion and potentially 10% to 12% of GDP in the next recession — deficits typically widen by 4 to 6 points in one — he says there are two exits and only two.
"You're going to inflate the deficit away by printing money or devaluing the deficit therefore or are you going to restructure the Treasury debt by extending maturities and or reducing coupons? Those are really the only two methods that you can do it." Cutting entitlements slows the accumulation but does not change the trajectory, because the debt already exists
The AI borrowers are price-insensitive, which makes the problem worse. "And meanwhile, the hyperscalers and the AI guys, they have an insatiable demand now for borrowing money at today's rates. The spreads went wider and they didn't care. And they won't care. They won't care if the rates go up 200 basis points."
His reason for expecting casualties is arithmetic: "They're hoping to get what did SpaceX say that they their addressable market is one quarter of global GDP. How many companies can have one quarter of global GDP as their market? Four."
7. The Energy Shock
Oil above $100 — 106 on the day, La Roche noted — leads into the part of the conversation with the most specific numbers.
The US Strategic Petroleum Reserve is nearly exhausted, and cannot be run to zero. "You can't deplete it all the way." Below a certain level the ratio of oil to the salt in the caverns compromises the oil; his energy team tells him the reserve is close to that point.
"Meanwhile, it's not just the United States. It's the global oil reserves are at basically the all-time low." The population is larger and the cushion is smaller.
"Diesel nationwide is eight a gallon now. And in California, there's some parts where it's $9.99 and 9/10." La Roche guessed the reason correctly before he said it: "They don't have enough digits on the pump." One suggested fix, which Gundlach thinks is workable, is to sell diesel by the half gallon and change a gear in the pump.
The hoarding has reached consumer products. "I just heard today on my way over here that Costco is rationing its Kirkland motor oil" — the price had roughly doubled, and the retailer was either seeing or anticipating stockpiling.
"This is what causes inflation problems when people start buying in advance because they think the price will go further up." He said the energy market is on the precipice of that, and La Roche added lubricants, helium and sulfur to the list; Gundlach added fertilizer.
"And I don't see this energy price shock going away at all."
8. CPI Starts With a 4
This is the call the interview turns on, and it is set directly against the Fed chair's own target.
"We think that the next print that comes out on the CPI will start with a four" — a forecast his firm's inflation model produced using last week's oil price, before the latest move, so he expects it to be worse with oil higher.
On the persistence: "based upon the oil movement and stay above four unless something changes with the commodity complex stays above four all the way through March." He was less confident about PCE, saying DoubleLine's model for it is not as robust.
Against that he set Warsh's stated definition of success. "And here's Kevin Warsh talking about we will get to two. And two doesn't mean 2 anything. He says two, not 2.5, two." Warsh has said the Fed will not fail, which Gundlach noted means failing by the chair's own definition if it does not get there.
"We think the long end continues to move higher if it's allowed to move by its own accord." And on the currency: "And the lower dollar isn't going to help inflation either."
9. The German Bund Model
La Roche asked him to explain a model he uses to locate fair value on the 10-year, and he explained it without defending it.
Nominal GDP used to be the benchmark for the 10-year yield, and specifically its 7-year moving average, which rises when the economy improves on a multi-year view.
Negative rates in Europe broke it. Foreign yields were dragging US yields lower, so US nominal GDP alone stopped explaining the 10-year.
The fix was to combine the 7-year average of US nominal GDP with the German 10-year yield, which brings the global interest rate picture into the signal. "And it turns out it's an amazing correlation." He put the R-squared at 0.93 over several decades, and said it is higher still if the first three years of the sample are dropped.
He labeled his own method: "I mean just admitted up front." This is data mining, he said, and it is still "a great starting point" because the moving average changes slowly enough that nobody needs to watch it daily.
What it says now is that the US 10-year is about 20 basis points too high — unusual, he said, because the two lines are normally almost on top of each other.
10. When Duration Gets Cheap
Asked what would make him extend maturity, Gundlach gave a number.
"I want a real interest rate of 2%."
"So if inflation is at two, I could maybe buy it at four, but inflation is more like at four. So I'd want to buy it at six in the present moment." And he added: "And I'm not convinced inflation is going to stop at four" — commodity prices are at a new high again.
The consumer evidence he cites is behavioral. "Consumers have moved down from, higherend retailers to middle-end retailers to lower-end retailers."
The indicator he found funniest is the gas station indicator. "When people go in and only buy gas, they don't get a Twinkie, they don't get a can of Coca-Cola because they don't want to spend the money on it." He conceded it is not the most scientific measure. "And so that kind of shows you that people are running out of money." The savings rate is very low as well.
He connected the rate level to the credit stress in section 3. Floating-rate bank loan borrowers of poor quality are, he said, playing beat the clock — trying to survive to the next rate-cutting cycle, which is what would bail them out.
11. Tomorrow's Meeting
The interview was recorded the day before the Fed decision, and Gundlach made a call.
"And that right now is a zero probability that they're going to enter into a rate cut cycle at tomorrow's meeting." He repeated it: "I think there's zero chance of a cut." He put the odds of a hike at not much better than even.
"If they didn't hike rates tomorrow, I think the 30-year Treasury would go up at least 20 basis points by the close of the day."
The two-year is the strongest argument for a hike. "Well, that's one of the reasons they're going to hike is the two-year is 100 basis points above the Fed funds rate right now, which is pretty far out there." That gap is about two-thirds as wide as it was in 2022, and on his reading the two-year is asking for 50 basis points. He expects 25.
Why he does not simply follow the market pricing: he cannot read Warsh yet. The Bloomberg probability function was around 88, and historically a Fed chair goes with the bond market above 70 in either direction. "His first press conference was almost 180 degrees different. The second one was almost 180 degrees different than the first." Gundlach compared it to Al Gore turning up as three different people across the 2000 presidential debates, so that nobody could work out what he was about.
What tipped him: Warsh's Jackson Hole speech opened with stories about hiking. "And then he's talking about he doesn't like forward guidance. He's not going to do forward guidance." Then he told an anecdote about a vigorous hike and a more relaxed one. "I'm going to go that he's going to hike 25."
12. Advice for Bessent
A listener asked what advice Gundlach would give the Treasury secretary. The answer was to resign while the getting is good.
"So, it's an impossible job because you're trying to do too many things at one time with one, with just a couple of policies at your disposal."
"You're trying to deal with a runaway deficit problem at a time when the Federal Reserve chair acknowledges that probably he's going to have to be hiking rates at a time when the president wants interest rates to be zero."
On the president's stated goal of having the lowest interest rates in the world: "We've got enormous debt. Why should we have the lowest interest rates? People aren't willing to lend us money at zero."
"The problems are very complicated and interconnected and, you got a hammer."
He also thought the public posture backfired. Bessent's line was "I am the house." Gundlach's verdict: "And what did everybody do? They bet against the house."
13. The Insurance Bomb
The longest single stretch of the interview is a warning about where the credit cycle ends, and it starts with a correction to a common trade.
"Most people don't understand that 30-year tips do not protect you from rising interest rates." Nominal and inflation-protected yields have moved together for six years and the difference has been constant. "So, rates went up 500 basis points on nominals. They also went up 500 basis points on tips." The price loss was the same.
"It's the ones that protect you from inflation are 5 years and in" — short-dated TIPS, not long ones.
On ratings capacity: one firm's affiliated insurers went from 3% of investments in affiliated assets to about 42%, and half of all their ratings came from a single agency with a 25-person staff that rated 3,200 deals in a year. "That's not possible. You have to read go through research packages of 200 pages to come up with a foundation for making ratings. They don't have the personnel to do it." The Justice Department is investigating a private credit firm over its ratings behavior.
His historical analogy is 2006, and the evidence is what happened to genuinely good paper. AAA-rated prime mortgage securities from Wells Fargo, on 80% loan-to-value loans to prime borrowers, traded below 30 in March 2009. "The AAA's went below 30 and they spent considerable length of time at prices below 60." "It turns out that they never lost any money. They've recovered all the way to par, but they clearly the market didn't perceive them as AAA."
"And so the life insurance industry has become a very dangerous place because they I don't think they have the money to pay all these long duration claims." Long-dated liabilities make private credit a reasonable asset for an insurer in principle, because illiquidity costs them nothing — but the firms selling it are now offering liquidity they may not have.
The ownership chain is the problem. "It's all a circle. It's all incestuous." The insurer is owned by a private credit firm owned by private equity, buys what it is told to buy, and reinsures offshore in Barbados or the Caymans where US regulators have no authority. Reported reserve buffers have fallen from $14 to $10 per $100 of future liability, and some, he said, may not hold the $100 at all.
Insurers also shop for jurisdictions the way borrowers shop for ratings, concentrating in a handful of states with looser oversight that are also the least troubled by money being pushed to an offshore reinsurer.
"So this is a very dangerous situation where the private credit is sort of the fuse and the insurance companies are the bomb."
His practical advice to individuals: "So, I would tell people if you're in the market for involved in life insurance or annuities, you should get it only from mutual companies because they work for the policy holder." A mutual is owned by its policyholders and makes its own investment decisions.
14. The Bailout Problem
The closing stretch is about what happens politically when the losses arrive.
His metaphor for the last several years is a gold-rush town. "And then they discover gold nearby. And suddenly the town 10 times the size at least." A meaningful fraction of the arrivals are opportunists trying to make a quick profit and leave before it stops. "Like I said in the past, it's the wild west." He dates it from about 2019.
Why the money went into private markets at all: in late 2021 bond yields were zero and stocks were expensive on traditional metrics, so investors chose vehicles precisely because they could not see inside them. Gundlach's version of the investor's reasoning is that if a strategy can be mapped back to stocks and bonds, the investor will refuse it — so the appeal was opacity.
On the fiscal precedent: "We gave we spewed out $7 trillion of giveaway money and we were too stupid to not think that inflation was going to go towards double digits."
He told the story of explaining this to his mother at five years old, in Buffalo, when the family could not afford a clothes dryer. He suggested the government give everyone a million dollars. Her answer: "Oh, well, the problem with that idea is an apple would cost $100."
"And we had J. Powell say inflation was transitory." A five-year-old understood it, he said, and the Fed chair did not
What worries him is that the template is now established, and the next bailout is harder to sell. A proposal of $5,000 per adult would cost about $1.25 trillion on a deficit already at 6% or 7% of GDP. "I think a lot of people would be livid if they bailed out these private credit people" — people who call themselves the most sophisticated investors on Wall Street. "They shouldn't be bailed out for taking sloppy protocols and risk management that's very poor." The 2008 bailout could at least be sold as protecting homeowners, even though he says it was really a bank rescue.
He offered a motive for the sudden AI-doom narrative, while disclaiming conspiracy thinking. Nuclear weapons were built by the defense department, not private industry, so that they could be controlled: "We didn't let private industry do that. That was the government." "But now if you create hyperbolic risks of AI, suddenly you have a good reason for the government to put in some sort of an oversight, which means of course picking winners and losers. And that is a very very unhealthy path."
Bonus Insights
Gundlach takes credit for killing an industry euphemism. After he criticized the term "semi-liquid" at the Milken conference in May, the industry dropped it that day. His objection: "Semi means half and liquid means you can get your money out." You can get your money out half the time, and it is the half when you do not want it.
He reached for Warren Buffett to describe what comes next: "You'll find out who's swimming naked when the tide goes out."
He dates the peak of the mood precisely. "This feels like 2006. This feels like 1999 where it's a can't lose mentality" — and he puts the top of the optimism at around June of this year. "Reminds you of 99 2006. We know how those movies played out."
The detail he could not explain: an executive who moved companies, resigned, and then warned that AI would kill everyone. "His ex page had been dormant and within hours it had 150 million views." He noted the timing relative to the midterms and to an IPO being organized for October, while saying he is not a conspiracy theorist.
The digression that produced the interview's only joke at his own expense was his recollection of George W. Bush's "Need some wood?" line in the 2000 debates. "And I own a timber ranch actually."
His closing line, and the source of the episode's title: "It's as I said six months ago, it's going to get harder during the year and I think we've crossed over to the hard side of the street for the next 6 months to nine months."
Gundlach's bottom line is that the losses from the AI build-out are the mechanism for the next major drawdown in risk assets, and that the damage will not stay inside technology: it runs through bought credit ratings into private credit and out the other side into life insurers and annuity writers, which is why he has moved his own recommendations to short duration, high credit quality, gold and commodities, and away from anything at the center of it.
Products, Companies & Tools Mentioned
DoubleLine Capital (His firm, and the source of the funds in the allocation: the total return fund with no corporate bonds, the rules-based commodity ETF up 38% year to date, the commercial real estate ETF yielding about 6% at a duration of two, and the flexible fund benchmarked against both cash and the bond index)
SpaceX (Rated BBB− on its bond issue while the bonds traded three notches lower, and the source of the addressable-market claim he says cannot be true for more than four companies)
Oracle (The other recent issue whose bonds widened sharply immediately after sale, which he reads as the market rejecting the rating)
Costco (Rationing Kirkland motor oil after roughly doubling the price, which he treats as the first consumer sign of hoarding-driven inflation)
Wells Fargo (The best mortgage underwriter of its era, whose AAA-rated prime securities still traded below 30 in March 2009 before recovering to par)
Books & Resources Mentioned
Gundlach Unlocked (His quarterly webcast, where the four-part allocation and the US nominal GDP versus German 10-year chart are published)
Watch the full episode:
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: