The AI bubble debate has officially moved past "is it a bubble?" to "how do I make money before it isn't."Speaking at the Greenwich Economic Forum this week, Galaxy Digital (GLXY) founder Mike Novogratz said that AI is in the "biggest bubble of our lifetime"... and that investors should jump right in. His reasoning, via Bloomberg, was a gem:"Bubbles don't end like we are today. I know how bubbles end, and they end spectacularly, and this isn't spectacular enough."In other words, the bubble is real, it's just not bubbly enough yet.Some 9,500 miles away in Singapore, Ray Dalio reached the opposite conclusion from the same premise, warning that AI is a "classic bubble" that is now close to bursting. Meanwhile, back in Greenwich, Nassim Taleb said both of them may be looking in the wrong place: the real fault line runs through the bond market. Below we lay out all three views, alongside Goldman's history of how every great capex bubble since the railways has ended (spoiler: not with a whimper, but not with a warning either), and why we have spent so much time over the past year on the debt holding up the AI supercycle."I Know How Bubbles End"Novogratz's bull case boils down to valuation: AI-linked stocks, he argued, still look cheap on a price-to-earnings basis. And for those still sitting on the sidelines, he had some practical advice: "If you were not invested in AI, you might as well just go home and put your head in a bucket of ice."He also sees no political brakes, saying both parties view the AI race with China as "somewhat existential," so Washington will have "a really hard time slowing AI down, even to the level that the guys building it want to slow it down." Of course, Mike is just talking his book here: the reality on the ground is different, with politicians on both sides approve the open-ended funneling of record amounts of debt into the AI bubble, (also called "growth") which is now soaking up more 10 year equivalents than the US government itself, they are dreading the accelerating animosity on the ground toward chatbots and their massive data centers, which most view as pushing electricity costs to record highs while setting the stage for the biggest layoff wave in history.Readers with long memories will also note that this isn't Novogratz's first "biggest bubble" call. Back in November 2017 he said that crypto would be "the biggest bubble of our lifetimes", also while staying bullish. Bitcoin peaked the following month and then lost more than 80% of its value, and Novogratz's own firm lost a staggering $273 million in 2018. In fairness, he was eventually proven right on bitcoin; it just took a spectacular crash first.Dalio: "I Think We're Close To That"Speaking in Singapore on Wednesday, the Bridgewater founder said that AI is a "classic bubble" nearing its bursting point, thanks to two forces: rising interest rates and the need to turn paper wealth into cash."We're in the part of the cycle that is before that but approaching that. I think we're close to that."Dalio - who has turned downright apocalyptic since severing all ties with his former hedge fund a few years back - has been making some version of this case since June, when he told Bloomberg TV that "the pricking is the converting of wealth into money". What has changed since then is the cost of money: long-dated Treasury yields are now at their highest since 2002, and Dalio added a second warning a day earlier, saying that China and Japan may pull back from US Treasuries, because a debtor-creditor relationship that is also an adversarial one "creates a very difficult dynamic" Incidentally, we pinned the relentless Treasury selling on Japan last week, before Japan moved on to France's OATs this week and sparking the latest European "doom loop", sending banking stocks plunging, so we'll allow ourselves a modest victory lap. Dalio's other observation is just as relevant (and also cribbed from us): AI firms that used to raise capital through equity "now have to shift to debt", just as we previewed last October. Which brings us to Taleb.Taleb: It's The Bonds, StupidThe Black Swan author agreed with Dalio's diagnosis but not his culprit. "The bond market is very vulnerable," Taleb said in Greenwich, according to Bloomberg. With the US forced to fund massive deficits from buyers who may lose their appetite, Taleb is less worried about Beijing or Tokyo than about domestic retirees and funds stepping away from long-duration debt, which he argues would be a far bigger blow.That is the same buyer base the AI complex is now competing for. As we put it last month, the record gap between AI-related investment grade bonds and the rest of IG is the heart of the problem:Unprecedented divergence between Investment Grade ex-AI and IG AI, which one year ago traded 20 tight to IG and now trades a record 50bps wide and increasingly like junk. This is the heart of the debt bubble, as it is also drawing demand away from Treasuries https://t.co/rAQdXeTH2I — zerohedge (@zerohedge) September 12, 2026Taleb also had a warning for the "buy AI because AI is changing the world" crowd: it would be "a trap" to buy the companies riding the boom just because the technology is transformative, since historically "the companies that benefitted were not the pioneers." He also reminded everyone that the S&P is not the economy: its biggest names are global businesses, and middle-income families are a better gauge of America's health than the index.True to form, Taleb wouldn't call a crash ("It's false to think in terms of the market will go down... You think in terms of risks increase"), and his advice was characteristically Universa-flavored: "Always have a tail hedge, even if you have no reason to hedge." (Shocking advice from a tail-hedge fund's in-house scientist.)What Goldman Says: "Normally The Trigger Is Always The Same"So who is right: the guy saying the bubble isn't spectacular enough to pop, or the guys saying it's about to? Goldman's Bobby Molavi, in his latest "Ruminations" note (available to pro subs), went through every great capex boom of the past two centuries and found that they all ended the same way, which isn't exactly Novogratz's "spectacularly":"This time is different...i'm sure....i hope....lets all hope....because.... Railways in 1845….Utilities in 1929….Japan in 1989…..Dotcom in 2000…..Shale in 2014…..normally the trigger all ways the same....a central bank pivot. Same order each time - stocks peak, capex keeps rising (for 6 to 24 months), then capex roll....then stocks roll."Put differently, by the time the bubble looks spectacular, the top is already in, and capex keeps rising for up to two years after the stocks have rolled over, which is exactly the phase in which everyone points to the still-rising spending and says "see, no bubble."Molavi also has the AI big 10 at 42% of S&P market cap, above the TMT peak of 41% in 1999 and second only to the 1830s railroad mania (63%) and Japan in 1985 (44%), "with chance of that going up depending on 1 or 2 new IPOs." Nothing says "not a bubble" quite like the AI trade overtaking the dotcom bubble on concentration:Incidentally, that's the same chart has been used frequently by Michael Hartnett over the past few months; not surprisingly, the BofA strategist is just as skeptical over how the AI "supercycle" ends, calling "AI the biggest bubble since the railroads".And the flip side of concentration is breadth, which Molavi shows has just plunged to the narrowest level since the dotcom peak:To be fair, Goldman isn't unanimous. Goldman's Tony Pasquariello, in his latest markets/macro note (also available to pro subs), effectively made Novogratz's valuation argument with data: the forward P/E of the median AI infrastructure stock has dropped from 32x in April to 22x, and these companies "are delivering, yet the market is approaching them with a degree of sobriety that was missing in years like 1999, 2007 or 2021." He also noted that AI infrastructure stocks are expected to drive half of S&P's Q3 earnings growth, with NVDA and MU alone driving a third of it. Diversification at its finest.The problem with "cheap on P/E" is that the "E" increasingly has to fund the capex, and not much is left over for the owners. As we noted last month, the free cash flow yield of the 10 largest stocks is now the lowest since the dotcom bubble:Elsewhere on the sell side, Bank of America sees a tech bubble brewing and advises clients to pile in anyway, using equity derivatives to hedge the fallout, while DBS CIO Hou Wey Fook argues that with Nvidia trading in the mid-teens on forward earnings, "how can it be a bubble?" Which, if you're keeping score, makes it two billionaires, one bank and a CIO who agree it's a bubble (or isn't) and that you should buy it either way.Bottom LineSo while Novogratz is waiting for "spectacular," Molavi's history says the trigger is far more boring: a central bank pivot. And that one has already arrived. As Pasquariello notes, the market now prices one more Fed hike this year and roughly two more next year, while long-end yields sit at 24-year highs, and Pasquariello himself admits that "when the Fed tightens liquidity, you need to take it seriously."Goldman's own gauge agrees: the "Monetary Policy Optimism" component of its Risk Appetite Indicator has just collapsed to -7.98, within a whisker of the all-time low of -9.38, even as the S&P sits at a record:Dalio's rates-and-liquidation mechanism and Taleb's fragile bond market are two sides of the same coin, and the AI debt binge sits right in the middle of it, crowding out Treasuries for the same duration buyers.So while it's possible that this bubble still has a spectacular blow-off ahead of it, the more likely outcome is that, as Molavi lays out, the stocks peak quietly while the capex keeps rising, and the "spectacular" part only becomes visible in hindsight. Then again, Novogratz did say it's the biggest bubble of our lifetime, and on that, at least, everyone agrees.Much more in the full Goldman "Ruminations", "markets / macro" and Bank of America's "Low breadth is a feature not a bug" notes, all available to pro subs.
"Not Spectacular Enough": Novogratz Says AI Is The Biggest Bubble Of Our Lifetime... And You Should Buy It
Full Article
Original Source
Read the full article at Zerohedge →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.