US Stock Short Positions Hit Record Highs as AI Risk Fuels Bearish Sentiment
Claire Weston
US short interest has climbed to the highest level since S3 Partners began tracking in 2010 — S&P 500 shorts near 3.79% of free float, NYSE shorts at 9%, surpassing both the financial crisis and the pandemic — with AI capex payoff now the market's central unresolved question.
How high has short interest actually climbed?
S&P 500 short interest is approaching 3.79% of free float — the highest since S3 Partners began tracking in 2010.
Russell 3000 short interest has risen to 6.3%, also a record; NYSE-listed stocks hit 9% by late June.
This means → current shorting intensity exceeds both the 5% peak during the 2008 financial crisis and the roughly 6% peak during COVID.
Brian Reynolds, chief market strategist at Reynolds Strategy, describes the recent build-up as "almost vertical."
What are the shorts betting against?
Ihor Dusaniwsky, head of predictive analytics at S3 Partners, notes that both the scale and the breadth of shorted names are expanding.
The core driver is doubt over returns on AI-related capital spending — the S&P 500 fell 1.6% last week on AI capex concerns and Chinese competition pressure.
In plain terms = the market isn't questioning AI itself — it's questioning whether the massive spending will actually earn its cost back.
With this much shorting, why hasn't the market collapsed?
S3 Partners data shows long positions are roughly twice the size of shorts; many bears are hedging their bets with net-long exposure.
This means → much of the short interest is part of long-short hedging structures, not naked directional bets — one-sided selling pressure is weaker than the headline numbers suggest.
Goldman Sachs data shows hedge funds have been covering US single-stock shorts at the fastest pace in three months.
Reynolds argues that elevated shorts and buying power are cancelling each other out, keeping the market range-bound for the past month — a process that may have already absorbed some speculative excess.
Which names are shorted the hardest?
Hertz Global Holdings has fallen 65% this year, with roughly 79% of its float sold short.
The most-shorted cohort in the Russell 3000 is down an average of 15% year-to-date, while the rest of the index is up about 21%.
This reflects a targeted, not blanket, bearish stance — shorts are concentrated in names with deteriorating fundamentals, and most of those bets have paid off.
The largest S&P 500 shorts by dollar value cluster in the "Magnificent Seven" and chip stocks such as Micron and Broadcom.
After the chip sell-off, what is Wall Street saying?
The VanEck Semiconductor ETF (SMH) dropped 8.9% last week; the Philadelphia Semiconductor Index (SOX) fell nearly 10% — the worst weekly decline since April 2025.
One catalyst: China's Moonshot AI released its Kimi K3 frontier model with partially open weights, pressuring US tech names.
Mizuho analyst Vijay Rakesh argues AI capex and compute build-out still have ample runway, driving semiconductor demand through 2028–2029 and beyond.
Morgan Stanley analyst Joseph Moore calls the recent sell-off "an attractive buying opportunity," naming Nvidia and Broadcom as the most compelling compute plays.
Is the correction over?
JPMorgan strategist Mislav Matejka writes that semis should soon attract buying support; meaningful new supply won't arrive until 2028, so "pricing in a turning point now is premature."
Evercore analyst Mark Lipacis is more cautious, characterizing the SOX's roughly 20% drop over four weeks as a "mid-cycle correction."
He expects the adjustment is not finished — "it may need another two to three weeks and a further 10–15% decline to complete."
In plain terms = the bulls see bargains already on the table; the bears say another leg down is needed — and the deciding factor is whether this AI earnings season delivers hard evidence that capex is converting into revenue.
Content is for reference only, not financial advice.