Key Takeaways

Rosenberg is willing to call it a bubble. "I don't mind calling it a bubble," he says, adding that "every bubble pops," though this one has not yet and bubbles "take time to fester."

The real bubble is investor behavior, not AI. He argues "the bubble is in investor behavior," pointing to leverage and sentiment: margin debt up about 50% in a year to roughly $1.5 trillion while the market rose 25%.

Sentiment and positioning are at extremes. He cites bullish sentiment near 78%, fund-manager cash around 1%, households at a record 73% in equities and just 7% in bonds, and a Shiller CAPE of 41, the highest since 2000.

His own strategy is built for capital preservation. Rosenberg describes a diversified, multi-asset approach (currently about 50% equities, 35% bonds and 15% hard assets) designed to "bunt for singles" and hedge the "rocket fuel" in a portfolio.

Labor could be the next surprise. He argues the Fed is ignoring a softening job market, and that sequential negative payrolls into the fourth quarter could flip the narrative from inflation back to recession.

Key Moments

00:17 - How Rosenberg is positioning for a fragile economy The strategy behind his defensive, multi-asset approach.

04:05 - Risk management, diversification and hard assets Why he treats diversification as a discipline, not a dirty word.

05:50 - Why Rosenberg likes 2-year Treasury notes The tactical Fed bet at the front of the curve.

07:55 - Is the AI boom becoming a bubble? Concentration, capex and the Nortel-Lucent lesson.

09:32 - "Every bubble pops": Rosenberg on AI excess Why he is comfortable using the word.

13:14 - Margin debt, extreme sentiment and record equity exposure The behavioral warning signs he watches.

14:59 - What's really driving Treasury yields higher? Real rates and hyperscaler capex, not fiscal fear.

22:30 - The labor market could be the next big surprise Why he thinks jobs, not inflation, drive the fourth quarter.

Is the Stock Market Going to Crash? Why David Rosenberg Says "Every Bubble Pops"

A market crash is a broad, sustained fall in prices, and with valuations near record highs, the question of whether one is coming is everywhere. Economist David Rosenberg, founder and president of Rosenberg Research, told Wealthion in September 2026 that he is comfortable calling this a bubble, that "every bubble pops," and that the dangerous excess is not the AI technology itself but investor behavior. He does not put a date on it, and this is his attributed analysis, not investment advice.

Is the stock market going to crash?

Rosenberg does not predict a date, but he is unusually direct about the setup. "I don't mind calling it a bubble," he says, and "every bubble pops." He is careful to add that this one "hasn't popped yet" and that bubbles "take time to fester," sometimes a year or more. He recalls being admonished at Merrill Lynch for calling housing a bubble in 2006 and 2007, and for flagging the dot-com bubble in the late 1990s, noting that in both cases the label was accepted only after the fact. His point is that the pattern is recognizable now, even if the timing is not. This is the same debate he takes up from the economy side in his companion interview on why the US economy is more fragile than it looks.

Is AI a bubble, or is the bubble somewhere else?

Rosenberg's key twist is that "the bubble is not in the technology, the bubble is in investor behavior." He places AI in a long line of genuine innovations, from railways and electricity to the Nifty Fifty and the internet, that nonetheless produced manias, and he warns that owning the wrong names can be ruinous even when the technology is real, recalling that investors who held Nortel and Lucent in 1999 did well right up until those companies collapsed. Citing Jeremy Grantham, Charles Mackay's "madness of crowds," and Bob Farrell's rules on sentiment, he argues the excess is a behavioral one. For contrasting views on AI, see Peter Boockvar on the AI spending boom cracking.

What warning signs is Rosenberg watching?

Leverage and sentiment, more than valuation. He notes margin debt has jumped about 50% in the past year to roughly $1.5 trillion, twice the market's 25% gain, which he calls "a lot of leverage behind this sentiment." He points to bullish sentiment near 78%, a level he says has never been higher; portfolio managers and mutual funds down to about 1% cash; and the US household sector at a record 73% in equities versus just 7% in bonds. Valuations are stretched too, with the Shiller CAPE at 41 for the first time since 2000, but his emphasis is behavioral: "the multiple is the heartbeat of confidence," and "valuations don't matter till they matter, and by the time they matter your head is sliced off."

How is Rosenberg positioned?

Defensively, as a hedge against the riskier parts of a portfolio. Rosenberg describes his approach, drawn from his firm's model portfolio (up about 55% since early 2023 at low-to-moderate risk, with a beta around 0.7 to a 60/40 mix and a Sharpe ratio above one), as a multi-asset strategy spanning currencies, commodities, bonds and equities, currently about 50% equities, 35% bonds and 15% hard assets. The aim, he says, is to "bunt for singles and hit the odd double," targeting high-single-digit risk-adjusted returns over a cycle rather than chasing 15%. He stresses diversification across both asset classes and geographies (his book is not US-centric, and he played AI indirectly through emerging Asia rather than owning a tech ETF), and a willingness to take profits and rebalance. (Disclosure: Rosenberg is the research advisor to, not the issuer of, an ETF based on this model portfolio; this is described as his strategy for context, not as a recommendation to buy any fund.)

Why does he like 2-year Treasury notes?

Because he sees the Fed eventually cutting, and the front of the curve benefiting. Rosenberg calls his two-year position a tactical "bet on the Fed" that will not be there in a year. He points to late 2018, when the Fed under Jay Powell raised rates and still had hikes penciled in for 2019, only to cut three times, rewarding anyone positioned at the front end. He also notes that after midterm elections in which the governing party loses ground, history shows the economy slows, inflation slows, and the Treasury market rallies about 80% of the time, which he expects again, calling the current front-end rally "the thin edge of the wedge." His fuller take on the yield picture appears alongside George Goncalves on stealth tightening.

What is really driving Treasury yields up?

Not fiscal fear, in Rosenberg's view. He points out that "we weren't talking about deficits back in February when the 10-year note was 3.97%," so a 60-basis-point rise since is not a sudden discovery of the debt. He attributes it instead to real rates and the term premium, including hyperscalers funding their capital spending in the bond market because they are free-cash-flow negative, which lifts real rates, though he notes their widening credit spreads suggest that cannot continue forever. He is critical of the Fed's communication under Chair Kevin Warsh, arguing Warsh needs to explain why he is not raising rates, while also conceding the post-Bernanke Fed "way overcommunicated" with dot plots that have outlived their usefulness. He adds that with six straight years of deficits above 5% of GDP and debt above 100% of GDP, it is "a miracle" the 10-year yields below 5%.

Is the US stock market too big to fail?

Effectively, Rosenberg thinks the Fed would step in, but only under certain conditions. He argues the Fed, even under Warsh, would respond to a destabilizing equity decline, but stresses the trigger that historically moves the Fed is not falling stocks alone but stress in the credit market. He points to late 2018, when equity weakness did not stop the Fed, but problems in the corporate bond market did. Warsh, he suggests, likely has "a longer fuse" and a lower strike price than Powell, Bernanke or Greenspan, so a decline would have to be disorderly, and tied to credit, to prompt a rescue.

Could the labor market trigger a recession?

This is what Rosenberg is watching most closely into the fourth quarter. He argues the Fed is not focused on the job market, and that the softening there is being overlooked: nominal wage growth is slowing "visibly," which he reads as more slack than a 4.1% unemployment rate suggests, and the broad U6 measure sits at 7.9% versus 7.0% before the pandemic, when the Fed funds rate was similar to today. He warns that the recent negative payroll print may not be a fluke, and that if the fourth quarter brings sequential negative payrolls alongside a rising unemployment rate, "AI or no AI," it will be hard to deny a recession, and the market narrative will "flip back to the labor market." As always on Wealthion, these are Rosenberg's attributed views, not advice.

What Investors Should Watch

• Margin debt and sentiment: the leverage (near $1.5 trillion) and record bullishness Rosenberg calls the real bubble.
• Cash and equity exposure: fund-manager cash near 1% and households at a record equity weight.
• The 2-year note and the Fed: his tactical front-end bet on eventual rate cuts.
• Hyperscaler credit spreads: widening spreads that he says limit how long capex-driven yields can rise.
• Nominal wage growth and U6: his gauges of hidden labor-market slack and recession risk.

FAQ

Is the stock market going to crash? David Rosenberg does not put a date on it, but says he is comfortable calling this a bubble and that "every bubble pops." He stresses it has not popped yet and that bubbles take time to fester, so he treats it as a recognizable pattern rather than a timing call.

Is AI a bubble? Rosenberg argues the bubble is in investor behavior rather than the AI technology itself, citing extreme leverage and sentiment. He compares it to past manias in real innovations like the internet and railways, warning that even real technology can ruin investors in the wrong names, as Nortel and Lucent did in 2000.

What are the biggest warning signs? He points to margin debt up about 50% in a year to roughly $1.5 trillion, bullish sentiment near a record 78%, fund cash near 1%, households at a record 73% in equities, and a Shiller CAPE of 41, the highest since 2000.

How is Rosenberg positioned? He describes a defensive, diversified, multi-asset strategy, currently about 50% equities, 35% bonds and 15% hard assets, built for capital preservation and high-single-digit risk-adjusted returns. He is the research advisor to, not the issuer of, an ETF based on it; this is context, not advice.

Could the labor market cause a recession? Rosenberg thinks it is the key risk into the fourth quarter. He watches slowing nominal wage growth and a U6 unemployment rate of 7.9%, and warns that sequential negative payrolls with a rising jobless rate would make a recession hard to deny.

Which expert and interview does this article reference? This article draws on Wealthion's interview with David Rosenberg, founder and president of Rosenberg Research: "Every Bubble Pops: Markets Aren't Ready."

Full Transcript

Speakers: Maggie Lake (Wealthion host) and David Rosenberg (founder and president, Rosenberg Research). ASR errors corrected (names, terms) and filler removed; meaning preserved. A mid-interview Wealthion membership message has been noted rather than reproduced.

David Rosenberg (cold open): I don't mind calling it a bubble. Every bubble pops. Margin debt has soared 50% in the past year to about $1.5 trillion. What is going to be the surprise heading into the fourth quarter of the year, Maggie, is going to be the labor market.

Maggie Lake: You have a global ETF launching, appropriately called Rosie. Given the scenario you've laid out, an economy weaker than people think, earnings held up by AI but weakness beyond that, a Fed that may cut without rates necessarily falling if uncertainty persists, what is your strategy?

David Rosenberg: Let me be clear: I personally am not launching the ETF. I teamed up with an ETF specialist, Corton Capital, and I am the research advisor. It is entirely premised on the views and themes of me and my team at Rosenberg Research, and it's nothing new, because the genesis was our Rosie model portfolio, which is on our website. It launches in the second or third week of September. I have about 2,300 subscribers in 40 countries who buy my research to become better investors, and they pushed me to do this so they can invest alongside me. The model portfolio, with a unit holder of one, David Rosenberg, started at the beginning of 2023 and is up 55%, with a beta around 0.7 to a 60/40 mix and a Sharpe ratio north of one. It's classified as low-to-moderate risk, because I'm the flag bearer of capital preservation. It's the hedge against all the rocket fuel in the rest of your portfolio. And yes, this perma bear's model portfolio has been my best investment of the past three years, so go figure. I have no cash and no shorts right now, and it's perfectly liquid, though that doesn't mean it can't hold cash or that we couldn't short; in the ETF we could short, but shorting takes real skill and high conviction, because if you're wrong the negative carry is punishing. When I worked at Gluskin Sheff, they had about $3 billion in long-short funds, and I sat with the portfolio managers day in and day out for 12 years, right next to the CIO's office, so I learned a lot about discipline, results and risk management.

Maggie Lake: What makes your strategy a good counter to the rocket fuel?

David Rosenberg: The mandate, and it's in the prospectus, is low-to-moderate risk, so you have to be mindful of beta, the Sharpe ratio, volatility and the standard deviations around performance. When we first launched it in 2023, we back-tested it across decades against bear markets, and it outperformed every time by at least 10 percentage points. It's really a risk-management tool with narrow bands: you won't hit a home run in a bull market, but you won't strike out in a bear market. It's designed to bunt for singles and hit the odd double, and it's truly multi-strat, currencies, commodities, bonds and equities. Right now, for all the bearishness, it's about 50% equities, 35% bonds and 15% hard assets like commodities, so it's diversified. Could we raise cash? Sure. But the premise is that it's derived from our highest-conviction themes; I've never written a report on the benefits of cash. Some ideas have a three-to-five-year shelf life, some one-to-three, and some are tactical, like the two-year notes we hold now, which is a bet on the Fed that won't be there a year from now. The one thing I preach is that diversification is not a dirty word, so it's diversified across asset classes and geographies; it is not US-centric. We decided to play AI differently, taking some off the table by being long emerging Asia. People say valuations don't matter until they matter, and by the time they matter your head's sliced off, so we focus on valuations, not as a timing tool, but to know whether the wind is in your face or at your back. It's a fund of funds, an ETF whose underlyings are ETFs, a passive vehicle managed actively; we're not scared to take profits or rebalance. The aim isn't gross returns but risk-adjusted returns, high single digits over a cycle, because we're not willing to take on that much risk at these multiples, though we're not constrained and could be 100% equities.

Maggie Lake: The AI sector is so over-concentrated and responsible for so much growth. Are you worried about a downturn there, the capex spend and the revenue outlook?

David Rosenberg: We haven't owned a single tech stock or a technology ETF. You could argue that through MSCI Asia we had indirect semiconductor exposure via Korea and Taiwan, but no tech ETF, and we're still up 55% over three-plus years. I don't cry over spilled milk. But remember, in 1999 if you owned Nortel and Lucent you were doing great, right up until they basically disappeared. [Membership message noted.] I'm not drawing exact parallels, but I don't mind calling it a bubble. I did a podcast last week with my old friend Rich Bernstein from Merrill, and we agreed we won't call it a bubble, we'll call it a massive excess, a giant suds. Jeremy Grantham says bubbles are two-sigma events or more; I won't debate his math. But we've seen this before. We don't want to call it a bubble now because it scares and upsets people. When I called it a housing bubble in 2006 and 2007, I was admonished at Merrill Lynch; clients would rather you call it something else. But now we call it the housing bubble after the fact, don't we? Back in the late 1990s, if you called it a tech or dot-com bubble, you were viewed as a heretic, and people would say, define a bubble for me. Now we don't mind calling it a dot-com bubble, after the fact. Bubbles take time to fester, but when they pop, they pop. This hasn't popped yet, but every bubble pops. It's like Gerald Ford in the mid-70s not wanting to call it a recession, so he called it a banana. Fine, we won't call it a bubble, we'll call it a banana. But it's following the pattern of every major shift on the innovation curve, from the wheelbarrow to electricity to the railways to the Nifty Fifty to the internet to housing. And the reason people are shamed for it is that the bubble is not in the technology, the bubble is in investor behavior. Why did Charles Mackay write about the madness of crowds, and why did Bob Farrell build sentiment and the herd mentality into his rules? It's not just valuations, though the Shiller CAPE just hit 41 for the first time since 2000. It's the leverage: margin debt has soared 50% in the past year to about $1.5 trillion, while the market's up 25%, so leverage is running at double the pace. Bullish sentiment is up to 78%, which has never happened before. Portfolio managers and mutual funds are down to 1% cash, which we've never seen. The US household sector has 73% of its portfolio in equities, never that high, and only 7% in bonds. So I like what's unloved and under-owned. The multiple is the heartbeat of confidence, and it drives performance over time far more than earnings growth does.

Maggie Lake: Some are concerned about the fiscal situation, a declining dollar and purchasing power, and argue bonds aren't a good investment in that scenario.

David Rosenberg: I started in the business in 1987, and all people talked about were the twin deficits; I've heard it for 40 years. We weren't talking about deficits back in February when the 10-year note was 3.97%. So you're telling me that in the past four months we suddenly discovered fiscal policy is a problem? That's not what drove yields up 60 basis points; it's not a new story. What is new is that the hyperscalers are ramping up capex and, because they're free-cash-flow negative, funding it in the capital markets, which lifts the real rate. But their credit-default swaps and spreads are widening, so there will be financial constraints; that won't go on in perpetuity. Separately, Kevin Warsh has to clean up his act and explain why he's not raising rates; he'll have another chance in September, or at Jackson Hole. If he doesn't want to communicate, then cancel the press conferences entirely, because there has to be a balance. I'd be the first to say the post-Bernanke Fed way overcommunicated; the dot plots were useful around 2012, to keep markets from pricing in tightening off a couple of good quarters, but that was 14 years ago and they've outlived their usefulness. And with six straight years of deficits above 5% of GDP and debt-to-GDP over 100%, and no QE, it's actually a miracle the 10-year yields below 5%. FDR never ran deficits like that during the New Deal. But this comes to an end on November 3rd, because we'll have a split government and fiscal gridlock for two years. And historically, after midterms in which the party in power loses control, 80% of the time the economy slows, 80% of the time inflation slows, and 80% of the time you get a big rally in the Treasury market. Maybe it's different this time, but I'd bet with high confidence on a rally in the front end of the curve, which is already happening. This is the thin edge of the wedge.

Maggie Lake: Do you think the US stock market is too big to fail?

David Rosenberg: In a sense. The Fed, even under Warsh, would respond to a destabilizing decline in equities, but it depends on how fast and how deep. In 2022 we had a cyclical bear market; just before ChatGPT launched in the fall of 2022, the market was down almost 30% while Powell kept raising rates, but it wasn't destabilizing, just a horrible year if you were long equities. What ultimately gets the Fed's attention is whether a stock-market decline coincides with problems in the credit market. Back in 2018, the falling stock market isn't what stopped the Fed; it was when we started having problems in the credit market. Warsh probably has a longer fuse and a lower strike price than Powell, Bernanke or Greenspan, but the Fed would respond to a destabilizing decline, all the more so because you won't get one absent a problem in credit, and that always brings the Fed to the rescue.

Maggie Lake: Looking to the fall, what's top of your radar?

David Rosenberg: From an interest-rate standpoint, the labor market. This Fed hasn't been paying attention to it. Powell at least explained himself, saying repeatedly there was no inflation coming out of the labor market, which anyone can interpret to mean that whatever cost shock we get eventually hits the wall in the labor market: real wages decline, real spending falls, and demand becomes the referee for inflation after the initial shock. Warsh said none of that; all he talked about was inflation being above target, over and over, without telling us how he'd get to target, which frustrated the bond market. I like to focus on what others aren't, so heading into the fourth quarter, it's the labor market. Right now people write off the roughly minus 23,000 nonfarm payroll print because the unemployment rate ticked down, and we're still averaging about 20,000 a month, which has a plus sign. But what if by October or November those arguments fall away and we start printing repeated negatives? AI or no AI, that will be hard to square with no recession, especially if the labor force finally stabilizes and the unemployment rate starts to rise. The unemployment rate is a lagging indicator anyway; the broad U6 measure of idle labor is stuck at 7.9%, versus 7.0% just before the pandemic, when the Fed funds rate was around 1.75%, roughly where it is today. And I watch nominal wage growth, because the truth is in the price: if the labor market were really tight, wage growth wouldn't be slowing, and it visibly is, which tells me there's more slack building than a 4.1% unemployment rate suggests. There's a good chance this July payroll print proves to be no fluke, the unemployment rate hooks up, and we start talking less about inflation and more about the labor market, which is what Powell focused on when they first started cutting a couple of years ago. So I think we flip back to the labor-market narrative.