\u201cIt\u2019ll happen slowly, and then all at once.\u201d<\/p>\n
That\u2019s how Jim Morrow, founder and chief investment officer of Callodine Capital,<\/a> describes the eventual \u2013 inevitable \u2013 unwinding of what he calls \u201cthe most crowded trade in history.\u201d<\/p>\n Of course, he isn\u2019t just paraphrasing Ernest Hemingway<\/a>\u2014he\u2019s talking about the AI race, and the trillion-dollar deals so overstretched they\u2019re better described as knots than trades. And he\u2019s not alone in sounding alarms.<\/p>\n<\/p>\n Michael Burry\u2014the investor of Big Short fame who famously predicted the 2008 housing collapse\u2014broke a two-year silence this week to say nearly the same thing<\/a>:\u00a0 that Big Tech\u2019s AI-era profits are built on \u201cone of the most common frauds in the modern era\u201d\u2014stretching the depreciation schedule (some, including Burry, would say cheating the depreciation schedule).<\/p>\n And it landed with extra weight: earlier this week, Burry quietly deregistered<\/a> his investing firm, Scion Asset Management, effectively stepping away from managing outside money or filing public disclosures. Some analysts interpreted the move as less of an ominous sign and more, as Bruno Schneller, managing director at Erlen Capital Management, told<\/a> CNBC, stepping away \u201cfrom a game he believes is fundamentally rigged.\u201d<\/p>\n \u201cOn to much better things,\u201d Burry hinted<\/a> on X<\/a>, with a new launch expected on November 25.<\/p>\n Freed from the obligations of reporting and client management, Burry returned to X with a message that cut straight through the current AI euphoria. To him, the boom in GPUs, data centers, and trillion-dollar AI bets isn\u2019t evidence of unstoppable growth; it\u2019s evidence of a financial cycle that looks increasingly distorted, increasingly crowded, and increasingly fragile.<\/p>\n Burry put numbers to it. In a post on X, the investor estimated that Big Tech will understate depreciation by $176 billion between 2026 and 2028, inflating reported profits by 26.9% at Oracle<\/a> and 20.8% at Meta<\/a>, to name two of his specific targets.<\/p>\n Meta did not respond to a request for comment. Oracle declined to comment.<\/p>\n \u201cHe\u2019s spot on,\u201d Morrow tells Fortune. Morrow has been making the arguments for months, warning that a \u201ctsunami of depreciation\u201d could quietly flatten Big Tech\u2019s AI profits. Behind the trillion-dollar boom in chips, data centers, and model training, he argues, lies a simple but powerful illusion: companies have quietly changed the length of time they account for their machines\u2014and their semiconductor chips\u2014wearing out and depreciating.<\/p>\n \u201cCompanies are very aware of it,\u201d Morrow claims. \u201cThey\u2019ve gone to great lengths to change their accounting and depreciation schedules to get ahead of it\u2014to effectively avoid all this capex hitting their income statements.\u201d<\/p>\n Burry\u2019s post<\/a> drew viral attention. Morrow\u2019s been making the case longer, but he thinks the sudden resonance means investors are finally waking up to something fundamental.<\/p>\n \u201cThese aren\u2019t small numbers\u2014they\u2019re huge. And the fact that someone like Burry is calling it out tells you people are starting to notice what\u2019s happening between the lines of the balance sheet.\u201d<\/p>\n Here\u2019s how depreciation works\u2014or doesn\u2019t. When tech giants like Microsoft<\/a>, Meta, and Oracle build AI data centers, they buy tens of billions of dollars in GPUs, servers, and cooling systems. Normally, those assets lose value fast, cutting into profits. But recently, Burry claims, many companies quietly extended how long they claim those machines last\u2014from roughly three years to as many as six.<\/p>\n That simple change lets them spread out their costs and report fatter earnings now.<\/p>\n \u201cHad they not made those changes,\u201d Morrow says, \u201ctheir earnings would be dramatically lower.\u201d<\/p>\n Meta\u2019s filings<\/a>, for one, seem to at least corroborate the directionality of Burry\u2019s and Morrow\u2019s claim. Until 2024, servers and network gear were depreciated over four to five years; effective January 2025, Meta said<\/a> they would \u201cextend the estimated useful lives\u201d of \u201ccertain servers and network assets\u201d to 5.5 years.<\/p>\n \u201cDepreciation expense on property and equipment,\u201d Meta wrote<\/a> in its annual filing for 2022, was \u201c$8.50 billion, $7.56 billion, and $6.39 billion for the years ended December 31, 2022, 2021, and 2020, respectively.\u201d<\/p>\n In other words: depreciation was a major cost already, and management clearly chose to stretch the timeline over which those costs are recognized. The policy change doesn\u2019t prove Burry\u2019s dollar totals across firms, but it markedly moves reported earnings in the direction he describes by lowering near-term depreciation expense and pushing more of it into later years.<\/p>\n Morrow argues that the timing makes little sense. As the pace of technological change accelerates\u2014Nvidia<\/a> now releases new chips every 12 to 18 months instead of every two years\u2014the hardware becomes obsolete faster, not slower.<\/p>\n It\u2019s like any old technology. Take laptops: imagine trying to run the newest version of Adobe<\/a> Premiere Pro on a 2018 MacBook. Sure, it might boot up, but it\u2019s going to overheat, lag, or crash, because it simply wasn\u2019t built for today\u2019s compute demands. Old chips are the same way: they don\u2019t stop working, but they quickly lose their economic value as newer, faster models render them functionally obsolete, Morrow argues.<\/p>\n To be sure, Morrow\u2019s expertise is in value investing and high dividend companies, as opposed to the trendier, high-growth tech stocks: in fact, he said he has no long positions in technology broadly. So, he benefits if Big Tech multiples compress or if the markets begin to reprice the costs buried in AI spending. Still, his critique aligns with growing unease from other analysts.<\/p>\n Richard Jarc, an analyst at Uncovered Alpha, has raised similar alarms about the mismatch between AI chip lifecycles and corporate accounting.<\/p>\n He has said<\/a> that the latest generation of GPUs wears out far faster than companies\u2019 amortization schedules suggest. While some point to the continued use of Nvidia\u2019s H100 chips<\/a>\u2014released three years ago\u2014as evidence of longer utility, Jarc says that\u2019s misleading.<\/p>\n Demand remains high largely because a handful of companies are subsidizing compute costs for end users<\/a>, a dynamic that depends on investor cash flow rather than fundamentals, Jarc argues. More importantly, Nvidia has now shifted from releasing new chips every 18\u201324 months to an annual cadence. In that context, Jarc said, treating GPUs as if they\u2019ll remain valuable for five or six years is unrealistic: their true economic life, he estimates, is closer to one or two.<\/p>\n The Economist in Sept. called<\/a> the delayed depreciation \u201cthe $4 trillion accounting puzzle at the heart of the AI cloud,\u201d noting that Microsoft, Alphabet<\/a>, Amazon<\/a>, Meta, and Oracle have each extended the useful life of their servers even as Nvidia shortens its chip cycle to one year.<\/p>\n By The Economist\u2019s estimates, if those assets were depreciated over three years instead of the longer timelines companies now assume, annual pre-tax profits would fall by $26 billion, roughly an 8% hit. A two-year schedule would double that loss, and if depreciation truly matched Nvidia\u2019s pace, the implied market value hit could reach $4 trillion.<\/p>\n Not everyone buys the doom loop. In a note to clients sent this week, Bank of America\u2019s semiconductor team argued that the market\u2019s sudden skepticism about AI capex is evidence that the trade is far less overcrowded than credits claim.<\/p>\n The recent selloff in megacap AI names, the team led by Vivek Arya wrote, was driven by \u201ccorrectable macro factors\u201d\u2014shutdown jitters, weak jobs data, tariff confusion, even misinterpreted OpenAI comments\u2014rather than any real deterioration in AI demand. In fact, the firm pointed to surging ancillary segments like memory and optical (up 14% last week), as well as Nvidia\u2019s disclosure of $500 billion-plus in 2025\u201326 data-center orders, as signs the underlying spending cycle remains \u201crobust.\u201d<\/p>\n But Morrow is most worried that investors are mistaking raw spending for real growth. The market, he argues, have stopped distinguishing between capital intensity and genuine productivity. The AI boom has pushed valuations from the realm of software multiples into something closer to industrial-scale infrastructure math. Building just one hyperscale, one-gigawatt data center\u2014enough to support a cutting-edge AI model\u2014can run around $50 billion, with the majority devoted to GPUs, followed by the buildings, cooling systems, and power infrastructure.<\/p>\n Reality check\u2014\u201cnone of these companies has ever managed a $50 billion project before,\u201d he says. \u201cNow they\u2019re trying to do fifty of them at once.\u201d<\/p>\n