Goldman Sachs put out a note this week that stopped me mid-scroll: “The median S&P 500 stock carries short interest of 3.0% of market cap, the highest level since late 2011.”

Given short selling has a theoretical unlimited loss. When aggregate short positioning reaches levels not seen since 1 year post the Global Financial Crises year, it might tell you something meaningful, at least about institutional sentiment.

So we pulled short interest data over the last 3 months for the largest 1,000 US stocks, to see what's actually happening beneath the surface. Here's what we found:

• 64% of stocks had short interest increases over the past 3 months. Total shares that sold short increased by nearly 1.4 billion shares, which is a 9.8% increase in aggregate short positions across the index.

Instead of shorts rising in a very concentrated way, which can be a thematic trade or narrative, this recent change reads more like a macro stance: a belief that equity prices may be broadly ahead of fundamentals.

However, there is a bias from a sector perspective.

• Information Technology leads the bearish shift. 73% of IT stocks have had short interest increases, led by something you may be able to guess: software and services, not chips.

  • Software: 85% of names (or 55 of them) had increasing shorts where levels were already elevated. These are enterprise software companies built on subscription models that spent the last decade looking like they had unbreakable moats. The top 5 names also tell the story: UiPath (PATH, now 27.7% short), Workday (WDAY, 12.7%), Salesforce (CRM, 8.8%), HubSpot (HUBS, 8.7%), and DocuSign (DOCU, 8.3%). 

  • IT Services: 88% of names have increasing shorts (or 17), the highest rate of any sub-group in tech. These are the outsourcing and professional services firms, whose business model is built on billable hours from human experts. These include EPAM Systems (+8.9 ppts of float), DXC Technology (+8.4 ppts), and Gartner (+6.4 ppts).  The thread connecting both is AI disruption risk. Bears appear to be betting that AI either commoditizes these companies' products (packaged software) or eliminates the labor cost structure underpinning their margins (IT services). Whether that thesis plays out over months or years is the debate, but the positioning says these stocks are priced for a world that may no longer exist. Meanwhile, the AI infrastructure buildout narrative is apparently strong enough to push bears out of hardware names even as they pile into software. Only 56% of semiconductor names had short interest increases, which is meaningfully below the broader tech average. And computer hardware is actually being covered. Only 25% of hardware names had short interest increases, the lowest rate in all of IT. For example, Super Micro Computer (SMCI) and Hewlett Packard Enterprise (HPE) are having positions unwind. 

• The one sector where bears are retreating is Energy. Bears are actually covering in this sector at a significantly higher rate than anywhere else. Oil refiners, integrated majors, and distributors are all seeing short positions unwind. Whether that's a geopolitical premium, an oil supply thesis, like we believe, or simple relative value against an expensive broader market.

The rest of the sector order, for context:

Another point worth noting from the chart above: the historical “defensives” have high short interest; Utilities are at 72%, Health Care at 70%, and Consumer Staples at 54%. Goldman notes short interest for the median Health Care stock has reached a 30-year high, with Utilities and Consumer Staples near record levels. Bears are building positions in defensives at historically extreme levels.

Turning to a name by name basis, the most shorted stocks by percentage of float skews toward companies where the bears have the loudest, most specific cases. What's notable here is among the 60 stocks with short interest above 15% of float, 78% of them saw short interest increase further. These have the most short interest right now:

Another note from the chart above that’s worth flagging: while short interest has been rising for the median S&P 500 and Nasdaq-100 stocks, it has actually been declining for small-caps. That said, the median Russell 2000 stock still carries roughly twice the short interest relative to market cap as the median S&P 500 or Nasdaq-100 stocks, so the small-cap universe remains structurally more shorted than large caps even as those positions ebb.

A few ways for investors to think about this:

Short interest is not a sell signal on its own. Some of the most spectacular rallies in market history happened in heavily shorted names, when the thesis broke and short sellers scrambled to cover. High short interest and a positive catalyst can be powerful.

But 10-year high aggregate positioning deserves respect. Investors are wrong all the time, but when they collectively reach a level of bearishness not seen in a decade, it's a signal worth taking seriously. These are not retail sentiment surveys but positions with real capital behind them.

The AI disruption trade is alive in the short book. If you own names in enterprise software or IT services, it's worth asking honestly whether you've thought through the AI commoditization risk the bears are pricing in. 

Energy as a contrarian play. The covering in energy names stands out. When the crowd is broadly short and one sector is being quietly abandoned by bears, it’s worth watching.

The bears are back, and at a scale we haven't seen in a decade. Whether they're right is the question markets will answer over the coming months. But knowing where the bets are and where they're retreating provides perspective as you think about your own positions. Note: Short interest data sourced from Factset, based on the largest 1,000 stocks as of May 24, 2026.