Episode Summary
Executive Summary: FT Alphaville editor Robin Wigglesworth explains that hedge funds are not yet making a broad short bet against AI’s biggest winners, but are increasingly probing weaker AI-linked companies, including Oracle and utilities investing heavily in data-center power. Rising technology borrowing adds to signs of froth. Short sellers may help reveal where a downturn could hit, but cannot reliably time it; core AI firms may prove more resilient than indebted adjacent players.
Main Topics: Where hedge funds are positioning against AI: Available positioning data suggests hedge funds are not broadly shorting dominant AI companies such as Nvidia, but are testing selected weaker companies and businesses linked to the AI buildout. Oracle and credit-market signals: Oracle is described as a comparatively vulnerable member of the AI hyperscaler group, with short interest rising and investors also using credit default swaps to bet on its creditworthiness weakening—not necessarily on default. Utilities and AI infrastructure risk: Utilities have attracted unusually high short interest as they invest heavily in power infrastructure for data centers. If AI-related demand disappoints, companies that borrowed to expand could be left with costly assets and debt. Technology borrowing and market-wide effects: The speakers discuss rapid growth in technology-sector debt issuance and data-center financing. Even profitable technology companies could reshape bond markets and compete with other sectors for investor capital. Why short selling cannot time a bubble: Shorting is risky because losses can theoretically be unlimited if a stock rises. Hedge funds may wait for a bubble to begin collapsing before acting, so their positions are more informative about potential weak points than the timing of a correction. Bloom Energy and the episode’s short ideas: For the closing Long/Short segment, Wigglesworth names Bloom Energy as his short, citing its high valuation relative to forecast profits and its AI-data-center association. Katie Martin jokingly shorts crypto maximalists; both stress this is not investment advice.
Key Arguments: Hedge-fund short positions are elevated relative to recent history, but the evidence points to selective bets around the edges of the AI trade rather than a wholesale wager against its largest, most profitable companies. The potential vulnerability may lie in AI-adjacent firms that borrow and spend heavily to become major players, rather than in established technology leaders with profitable businesses and substantial cash flow. Utilities’ exposure is a possible indirect way to bet against AI: their investment in power supply could become a liability if data-center demand does not meet expectations. Technology borrowing is one sign of AI-related exuberance, and heavy issuance could affect other bond-market borrowers by increasing competition for investor capital. Short sellers should not be treated as reliable predictors of when a market correction will happen. Shorting a rising market is costly and dangerous, and hedge funds often move together as a herd. A transformative technology can still produce a damaging bubble: the speakers use the dot-com boom as an example of an investment thesis that could be directionally right while investors still suffer severe losses.
Data Points: S&P 500 value represented by short positions: About 2.4% - Katie Martin cites the share of the overall index value currently short, describing it as high relative to the past five years and above the average since 1995. Technology share of U.S. investment-grade corporate debt issuance: Up 80% this year - Cited from a UBS note as evidence of increased technology-sector borrowing. Technology-company bond issuance: Over $300 billion this year - Wigglesworth refers to major technology companies issuing this amount, primarily to finance data centers, while noting it is manageable relative to their cash flow. AI data-center and project-finance deals: Increased tenfold to at least $125 billion this year - A UBS estimate discussed by Martin, illustrating the scale-up in financing tied to AI infrastructure. U.S. utilities short interest: Highest since at least 1995 - Wigglesworth says Goldman Sachs data show unprecedented short interest over the period covered, as utilities invest to supply data-center power. Bloom Energy forecast earnings and valuation: About $100 million forecast earnings next year; around $30 billion valuation - Wigglesworth’s rationale for naming Bloom Energy as his Long/Short segment pick to short.
Pivotal Quotes: "You don't get run over. You wait until it actually is already collapsing and then you pounce on it, essentially." — Robin Wigglesworth: He explains why hedge funds may avoid shorting a bubble while it is still expanding and why their positions do not necessarily predict the timing of a downturn. "I think we shouldn't look to them to understand the timing of when the AI bubble might burst, but more to understand how it might start playing out." — Robin Wigglesworth: His central point about how listeners should interpret hedge-fund positioning. "Everything that people said in 1999 was completely right. But people still lost an ungodly amount of money in the years that came afterwards." — Robin Wigglesworth: He warns that a technology can be genuinely transformative while investment prices and investor outcomes are still badly distorted.
Implications: For investors, hedge-fund shorts may flag fragile parts of the AI supply chain but offer little guidance on when a selloff will begin. Watch debt-funded infrastructure and AI-adjacent firms as well as headline technology stocks; high short interest is not, by itself, proof of an imminent crash.
From the Episode
Starts. The reality is that hedge funds, at least the smart ones, don't want to bet against a bubble whilst it's still inflating. You don't get run over. You wait until it actually is already collapsing and then you pounce on it, essentially. I mean, most hedge funds are still YOLOing into technology stocks and AI stocks in general. They are still massively long NVIDIA and all the other AI companies. And that hasn't really changed. What is changing is that you can see. They're starting to map out a playbook of what they might start doing when eventually things do crumble. Because we know they always do. And it might be next week, it might be next month, it might be next year, or maybe in five years. We don't know. But you can see that where they think the most carnage is going to fold are all these sort of AI-adjacent sectors like energy companies, utilities that are kind of frankly probably borrowing, investing too much money to service something that might.
Because frankly, everybody else is still richer than us. Yeah, yeah. But it does feel right now that things have gone a little bit nutty around AI. And even if you believe that it is transformational, even if it does prove transformational, crisis can still be stupid. Yes. And people forget that dot-com, everything was right. Everything that people said in 1999 was completely right. But people still lost an ungodly amount of money in the years that came afterwards. Yeah. So, in that sense, it is a little bit of a warning sign that people are starting to put their money where their mouth is and bet against some of the kind of, you know, shakier bits around the edges of the AI trade. I mean, do you have a good sense of whether it makes sense to watch what hedge funds are doing? Are they the smartest people in the room and do they tend to move first in terms of picking market direction or are they just as lost as the rest of us? It's a good question.
happens. Yeah, there's a reason why we're financial journalists and not bazillionaires sitting on a Caribbean island by ourselves counting our massive stacks of money. Quite often people will point out that we are irredeemably cynical and skeptical of everything. And we were saying, where's the crash since I think the first time 2013? The advantage of being a journalist that we have no skin in the game. We'll be proven right at some point. And then we'll try not to be too smug about it. Because frankly, everybody else is still richer than us. Yeah, yeah. But it does feel right now that things have gone a little bit nutty around AI. And even if you believe that it is transformational, even if it does prove transformational, crisis can still be stupid. Yes. And people forget that dot-com, everything was right. Everything that people said in 1999 was completely right. But people still lost an ungodly amount of money in the years that came afterwards. Yeah.
About Unhedged
Katie Martin, Robert Armstrong and other markets nerds at the Financial Times explain the big ideas behind what’s happening in finance right now. Every Tuesday and Thursday. Hosted on Acast. See acast.com/privacy for more information.