A top Wall Street analyst has sounded an alarm over the U.S. equity bull market, warning that its remarkable run is built on a precariously narrow foundation: a surge in spending on, and optimistic assumptions about, infrastructure for artificial intelligence (AI). This spending has fueled a boom in the shares of most of the so-called Magnificent 7 and a few dozen related businesses, which have now come to account for roughly 75% of the S&P 500\u2019s returns since the rally of the last few years began.<\/p>\n
The commentary on September 29 by Morgan Stanley<\/a> Wealth Management\u2019s chief investment officer, Lisa Shalett<\/a>, frames the current market boom as a \u201cone-note narrative\u201d almost entirely dependent on massive capital expenditures in generative AI, raising questions about its durability as economic and competitive risks start to mount. Shalett\u2019s critique came squarely in the middle of some people in the AI field \u2014 and many financial commentators<\/a> around Wall Street<\/a> \u2014fretting at market exuberance and beginning to talk openly about a bubble<\/a>.<\/p>\n In an interview with Fortune, Shalett said she was \u201cvery concerned\u201d about this theme in markets, saying her office had broadened from a belief that the market would only bid up seven or 10 stocks to roughly 40. \u201cAt the end of the day \u2026 this is not going to be pretty\u201d if and when the generative AI capital expenditure story falters, she said. <\/p>\n Shalett said she\u2019s worried about a \u201cCisco moment\u201d like when the dotcom bubble burst in 2000, referring to the company that was briefly the most valuable company in the world before an 80% stock plunge<\/a>. When asked how close we are to such a moment, Shalett said probably not in the next nine months, but very possibly in the next 24. When you look at the actual spending and the amount of capital coming into the space, \u201cwe\u2019re a lot closer to the seventh\u00a0inning than the first or second inning,\u201d she said.<\/p>\n \u2018Starting to do what all ultimate bad actors do\u2019<\/p>\n Shalett\u2019s comments centered on several recent multibillion-dollar deals to scale up data-center infrastructure. As notable substacker and former Atlantic writer Derek Thompson recently noted in a post titled \u201cThis is how the AI bubble will pop<\/a>,\u201d so much money is being spent to support AI\u2019s energy-consumption needs that it\u2019s the equivalent of a new Apollo space mission every 10 months. (Tech companies are spending roughly $400 billion this year alone on data-center infrastructure, while the Apollo program allocated about $300 billion in today\u2019s dollars to get to the moon from the 1960s to the \u201970s.)<\/p>\n What\u2019s more than a little concerning to Shalett is that one company alone, Nvidia\u2014the most valuable company in the history of the world, with an over $4.5 trillion market cap\u2014is at the center of a significant number of these deals. In September alone, Nvidia invested $100 billion in OpenAI<\/a> in a massive deal, just days after pledging $5 billion to Intel<\/a> (the Intel<\/a> agreement was tied to chips, not data-center infrastructure, per se).<\/p>\n Fortune\u2018s Jeremy Kahn reported<\/a> in late September on significant concerns about \u201ccircular\u201d financing, or Nvidia\u2019s cash essentially being recycled throughout the AI industry. Shalett sees this as a major concern and a major sign that the business cycle is headed toward some kind of endgame<\/a>. \u201cThe guy at the epicenter, Nvidia, is basically starting to do what all ultimate bad actors do in the final inning, which is extending financing, they\u2019re buying their investors.\u201d <\/p>\n Shalett expanded on her concerns by saying that companies around Nvidia \u201care starting to become interwoven.\u201d She noted that OpenAI is partially owned by Microsoft<\/a>, but now Nvidia has also made an investment in the startup, while Oracle<\/a> and AMD each have their own purchasing agreements with OpenAI. But OpenAI also has a data-center deal with tech giant Oracle, with the \u201cbad news,\u201d Shalett notes, that this deal is \u201ctotally debt-financed.\u201d OpenAI also struck a deal in October with chip-maker AMD that allows OpenAI to buy up to 10% of AMD. \u201cEssentially, Nvidia\u2019s main competitor is going to be partially owned by OpenAI, which is partially owned by Nvidia. So, Nvidia can \u2018own\u2019 a piece of its largest competitor. It is totally circular and increases systemic risk.\u201d<\/p>\n When reached for comment, a spokesperson for Nvidia said, \u201cWe do not require any of the companies we invest\u00a0in to use Nvidia technology.\u201d<\/p>\n Nvidia CEO Jensen Huang discussed the OpenAI investment in an appearance on the Bg2 podcast<\/a> with Brad Gerstner and Clark Tang on September 25, calling it an \u201copportunity to invest\u201d and part of a partnership geared toward helping OpenAI build their own AI infrastructure. When asked about the allegation of circular financing in general and the Cisco precedent in particular, Huang talked about how OpenAI will fund the deal, arguing that it will have to be funded by OpenAI\u2019s future revenues, or \u201cofftake,\u201d which he pointed out are \u201cgrowing exponentially,\u201d and by its future capital, whether it\u2019s raised by a sale of equity or debt. That will depends on investors\u2019 confidence in OpenAI, he said, and beyond that, it\u2019s \u201ctheir company, it\u2019s not my business. And of course, we have to stay very close to them to make sure that we build in support of their continued growth.\u201d<\/p>\n Shalett said that she and her team were \u201cstarting to watch\u201d for signs of a bubble popping, highlighting the deal announced roughly a week before OpenAI struck its $100 billion data-center <\/strong>deal with Nvidia, when it struck another with Oracle<\/a> worth $300 billion. Analysts at KeyBanc Capital Markets estimated <\/a>that Oracle will have to borrow $100 billion of that amount\u2014$25 billion a year for the next four years.<\/p>\n \u201cEvery morning the opening screen on my Bloomberg is what\u2019s going on with CDS spreads on Oracle debt,\u201d Shalett said, referring to credit default swaps, the financial instrument that was obscure before the Great Financial Crisis, but infamous for the role it played in a global market meltdown. CDSs essentially serve as insurance to investors in case of insolvency by a market entity. \u201cIf people start getting worried about Oracle\u2019s ability to pay,\u201d Shalett said, \u201cthat\u2019s gonna be an early indication to us that people are getting nervous.\u201d She added that all the indications to her speak of the end of a cycle and history is littered with cautionary tales from such times. <\/strong><\/p>\n Oracle did not respond to requests for comment.<\/p>\n 90% growth since the last bear market<\/p>\n Since the October 2022 bear market bottom and the launch of ChatGPT, according to Shalett\u2019s calculations, the S&P 500 has soared 90%, but most of these gains have come from a small group of stocks. The so-called \u201cMagnificent Seven\u201d\u2014including high-profile names like Nvidia<\/a> and Microsoft\u2014plus another 34 AI data-center ecosystem companies, are responsible for, as cited by Shalett and separately by JP Morgan Asset Management\u2019s\u00a0Michael Cembalest<\/a>, about three-quarters of overall market returns, 80% of earnings growth, and a staggering 90% of capital spending growth in the index. Comparatively, the other 493 names in the S&P 500 are up just 25%\u2014showing just how concentrated the rally has become.<\/p>\n The so-called \u201chyperscaler\u201d companies alone are now spending close to $400 billion annually on capex supporting AI infrastructure, Morgan Stanley Wealth Management calculated. The economic influence of AI capex is now immense, contributing an estimated 100 basis points\u2014fully one percentage point\u2014to second-quarter GDP growth, according to Morgan Stanley\u2019s research. This pace outstrips the rate of underlying consumer spending growth by tenfold, underscoring its centrality to both market performance and broader economic data.<\/p>\n \u201cPeople conflate AI adoption, which is in the first inning, with the capex infrastructure buildout, which has been going full-out since 2022,\u201d Shalett told Fortune. She cited concerns about the prominence of private equity and debt capital coming into play, as that \u201ctends to produce bubbles, because it may be unspoken-for capacity.\u201d In other words, people have money to burn and they\u2019re throwing it at things that may not pay off.<\/p>\n Shalett waved away macro theories about the labor market or the Federal Reserve. \u201cWe think that\u2019s missing the forest for the trees because the forest is entirely rooted in this one story\u201d about AI infrastructure. Morgan Stanley\u2019s bull-case mid-2026 price target for the S&P 500 is an eye-popping 7,200, but Shalett highlights that even the most optimistic outlook admits that risk premiums, credit spreads, and market volatility do not seem to fully account for the vulnerabilities lurking beneath the AI-fueled advance. <\/p>\n Shalett\u2019s analysis suggests that AI capex maturity is approaching and some possible slowdowns are already visible. For instance, hyperscalers have already seen free-cash-flow growth turn negative, a sign that investment may have outpaced underlying technology returns. Strategas, an independent research firm, estimates that hyperscaler free cash flow is set to shrink by more than 16% over the next 12 months, putting pressure on lofty valuations and forcing investors to demand more discipline in how these funds are deployed.<\/p>\n Shalett was asked about data centers\u2019 disproportionate impact on GDP throughout 2025, which media blogger Rusty Foster of\u00a0Today in Tabs<\/a> described as: \u201cOur economy might just be three AI data centers in a trench coat.\u201d The Morgan Stanley exec said \u201cThat\u2019s what makes this cycle so fragile,\u201d adding that at some point, \u201cwe\u2019re not gonna be building any data centers for a while.\u201d After that, it\u2019s just a question of whether you crash: \u201cDo you have a mild 1991-92-style recession or does it really become bad?\u201d<\/p>\n A more bullish case<\/p>\n Bank of America<\/a> Research weighed in on the semiconductors sector in a Friday note, writing that vendor financing in the space, especially Nvidia\u2019s $100 billion commitment to OpenAI, has been \u201craising eyebrows.\u201d Nevertheless, the team, led by senior analyst Vivek Arya, argued that the deal is structured by performance and competitive need, rather than pure speculative frenzy. <\/p>\n In an interview with Fortune, Arya explained why he wasn\u2019t worried despite the \u201coptics\u201d being pretty obviously bad. \u201cIt\u2019s very easy to say, \u2018Oh, Nvidia is giving [OpenAI] money and they are buying chips with that money\u201d and so on, but he argued the headlines are misleading about how much money is actually being spent and the $100 billion sticker price on the OpenAI deal \u201cscared everyone.\u201d Noting that the deal has multiple tranches that will play out over several years to come, he said it\u2019s not like Nvidia is \u201cjust handing a $100 billion check to OpenAI [and saying] you know, go have fun.\u201d<\/p>\n \u201cNvidia didn\u2019t fund all of it,\u201d Arya said of the wider generative AI capex boom. Citing public filings, Arya argued that Nvidia\u2019s entire investment in the AI ecosystem is in fact less than $8 billion or so over the last 12 months, not such a large figure after all. And he\u2019s still bullish on Nvidia and OpenAI, he added, because he sees them as the winners of this particular story. \u201cWe think they are going to be among the four or five ecosystems that come up. It\u2019s not like Nvidia is going and investing in every one of those ecosystems, right? They\u2019re only investing in one of those five, which is, of course, the most disruptive,\u201d that being OpenAI.<\/p>\n When asked about his own fears of a bubble, Arya actually sounded a calmer but strikingly similar tune to Shalett. \u201cI\u2019m extremely comfortable with what will happen in the next 12 months,\u201d Arya said, \u201cAnd I have high sense of optimism about what will happen in the next five years. But can there be periods of digestion in between? Yeah.\u201d Explaining that this is the nature of any infrastructure cycle, \u201cit\u2019s not always up and to the right.\u201d In other words, after the next nine months in Shalett\u2019s opinion and the next year in Arya\u2019s, the data-center buildout endgame could be in play. \u201cWhen these data centers are built,\u201d Arya said, \u201cthey are not built for today\u2019s demand. They\u2019re built with some anticipation of demand that will develop in the next, you know, 12 to 18 months. So, are they going to be 100% utilized all the time? No.\u201d<\/p>\n Rising worries about a bubble<\/p>\n Some of the biggest names in tech and Wall Street offered were hedging hard about the possibility of a bubble on Friday. Goldman Sachs CEO David Solomon and Jeff Bezos, both speaking at a tech conference in Turin, Italy, said they were seeing the same patterns as Shalett. Solomon said<\/a> the massive amounts of spending weren\u2019t fundamentally different from other booms and busts. \u201cThere will be a lot of capital that was deployed that didn\u2019t deliver returns,\u201d he said. That\u2019s no different from how investment works. \u201cWe just don\u2019t know how that will play out.\u201d<\/p>\n Bezos characterized it as \u201ckind of an industrial bubble,\u201d arguing that the infrastructure would pay off for many years to come.<\/p>\n OpenAI CEO Sam Altman, who got markets jittery in late August when he mentioned the B-word, was asked again to comment on the subject<\/a> while touring (what else?) a giant new data center in Texas. \u201cBetween the 10 years we\u2019ve already been operating and the many decades ahead of us, there will be booms and busts,\u201d Altman said. \u201cPeople will overinvest and lose money, and underinvest and lose a lot of revenue.\u201d<\/p>\n