Hedge funds dumped single stocks in record numbers last week, with the bulk of the selling concentrated in technology and semiconductors, according to Bank of America data released ahead of Nvidia’s after-the-close earnings report on Wednesday, May 20, 2026. The aggressive de-grossing, totaling $4.6 billion in single-stock sales and $3.01 billion in tech alone, marks the largest weekly outright sale of single names ever recorded by the bank’s prime brokerage. The move suggests that smart money is positioning defensively even as retail investors continue to chase the artificial intelligence trade, and it sets up Nvidia’s report as one of the most consequential earnings events of the year. The story was first reported by CNBC’s Fred Imbert.
The scale of the selling is the headline. Bank of America tracks prime brokerage flows across its hedge fund client base in granular detail, and the report indicates that the dollar value of outright single-stock sales last week was unprecedented in the firm’s history. To put that in perspective, hedge funds typically rotate between sectors rather than reducing gross exposure outright. A simultaneous reduction in long positions and increase in short positions across the technology sector at this magnitude is not a normal rebalancing. It is a deliberate risk reduction ahead of an earnings event that the fast-money community has decided it does not want to be holding through, no matter how strong the print.
Tech and Chip Stocks Bore the Brunt
The PHLX Semiconductor Index, the benchmark for the chip sector, fell 4 percent on Friday, May 15, in its worst single day since March 30. The selling continued on Monday, May 18, with the index dropping another 2.5 percent. The two-day move erased weeks of gains in some of the most heavily owned semiconductor names and confirmed for traders that the chip rally that had carried the major indexes higher for most of the spring was running into resistance.
The hedge fund flow data helps explain the magnitude of the move. When the entire prime brokerage complex is selling the same sector at the same time, the orderly market mechanics that normally absorb supply break down. Index funds, retail brokerages, and corporate buybacks become the buyers of last resort, and prices fall faster than fundamentals would suggest they should. The chip sector’s price action over Friday and Monday looked exactly like that kind of forced selling, with the worst declines concentrated in the names hedge funds had previously held in the largest size.
Nvidia, which has been the single most important stock in the global equity market for the better part of three years, bore a disproportionate share of the selling. The company’s market capitalization remains the largest in the world, and any change in hedge fund positioning around the name moves the index. Goldman Sachs and Morgan Stanley both noted in client communications that the volatility in Nvidia options markets ahead of the print was at levels usually reserved for individual biotech binary events, not for a $4 trillion company with stable cash flows.
The Setup for Nvidia’s Earnings
Expectations for Nvidia’s quarterly report are extraordinary. Analysts polled by LSEG forecast that the company’s bottom line more than doubled from the year-earlier period, and the consensus revenue forecast calls for a nearly 80 percent year-over-year increase. Numbers of that magnitude would be considered transformational for any other company. For Nvidia, they are the floor, not the ceiling, and the market’s reaction will hinge on what the company says about the next quarter, the next year, and the durability of artificial intelligence capital expenditure broadly.
John Belton, portfolio manager at Gabelli Funds, framed the dynamic clearly in remarks to CNBC. “There’s a short-term, long-term dynamic, where in the short-term, I think the numbers will be amazingly boring as usual,” Belton said. “Nvidia has been consistently blowing away numbers, especially the last couple of earnings. The market has traded with jitters around Nvidia because the long-term is the question.” That question, Belton noted, comes down to whether the trillions of dollars in artificial intelligence capital expenditure flowing through hyperscalers, sovereign AI projects, and enterprise data centers is actually generating returns commensurate with the investment.
The Belton framing is the right one. The financial community has accepted that Nvidia will continue to grow revenues and earnings at extraordinary rates for at least another four to six quarters because the order book is locked in and the supply chain is committed. What it is questioning is the second derivative. Will the customers buying these chips actually monetize them at scale? Or is the AI capital expenditure cycle going to follow the historical pattern of capex booms, in which a small number of winners emerge and a large number of buyers find themselves with stranded assets? Hedge funds, by their nature, are paid to ask this kind of question early and to position before the consensus catches up. The $3 billion of tech selling last week is the prime brokerage manifestation of that question being asked at scale.
Why This Selloff Looks Different
There have been several rotations out of technology over the past 24 months. Some were triggered by interest rate concerns, some by geopolitical events, some by the periodic reassessment of valuation multiples. What makes the current move different is the combination of three factors that did not previously align.
First, the magnitude. Single-week sales of $4.6 billion in single names against a baseline of net buying for most of the past 18 months is a large move in absolute terms and an extreme outlier in statistical terms.
Second, the concentration. Two-thirds of the selling was in technology, with the remainder spread across consumer discretionary and communication services. That is a sector call, not a market-wide de-risking. Hedge funds were not selling everything. They were selling the AI trade specifically.
Third, the timing. The move occurred in the trading week leading up to Nvidia’s earnings, which is exactly the window in which positioning is most informative. Funds that want to be flat into a binary event sell in the prior week. Funds that want to be long sell early in the prior month and add back into the print. The pattern of the past week is consistent with a coordinated decision across the hedge fund community to be flat or short into the report.
These factors together suggest that the hedge fund community has concluded the risk-reward into Nvidia’s print is asymmetric to the downside, even with the underlying business performing well. Their thinking is that a beat is already priced in and the upside surprise scenario is therefore limited, while a miss or a softer-than-expected outlook would cascade across the entire AI complex. That kind of asymmetric setup is exactly when professionals reduce exposure.
The Broader Market Context
The hedge fund selling does not happen in isolation. The S&P 500 has continued to climb to new highs through May, supported by strong earnings from the largest technology companies and by continued retail participation. But the breadth of the rally has narrowed. Equal-weight indexes have underperformed market-cap-weighted indexes by a widening margin over the past month, and the percentage of S&P 500 components trading above their 50-day moving average has fallen even as the index itself has risen.
That kind of narrowing breadth is a classic late-cycle market signal. It tells you that the rally is being driven by a handful of mega-cap winners, that the average stock is no longer participating, and that any disruption to the leadership cohort will have an outsized effect on the index. Nvidia is the largest component of that leadership cohort by a wide margin, which is why its earnings have become a market-wide rather than a single-stock event.
Sarah Min at Bank of America noted in a related research piece that investors have been “flooding out of cash to buy stocks,” and that the rate at which cash is being deployed into equities now matches levels seen at previous market tops. Bank of America’s framework treats this kind of “bull capitulation” as a sell signal, on the view that cash flowing into stocks at this pace tends to mark the moment when there is no incremental buyer left to absorb supply at higher prices. The hedge fund selling last week is consistent with that view. The smart money is exiting while the retail flow is still strong enough to absorb it.
Sector Implications Beyond Chips
The selling has been concentrated in chips, but the implications run wider. The artificial intelligence trade has pulled in adjacent sectors over the past two years, including data center real estate investment trusts, electrical utilities exposed to AI demand, copper miners feeding the cabling buildout, and software companies positioning themselves as AI beneficiaries. Each of these adjacent trades has been priced based on continued strength in the underlying AI capital expenditure cycle.
If Nvidia’s report or its guidance suggests that the cycle is plateauing rather than accelerating, the adjacent trades will reprice along with the chips. Copper, in particular, has been trading at record highs partly on AI demand projections, and a softening in those projections would remove a meaningful portion of the demand thesis. The same logic applies to the utilities, where regulated returns are not at risk but where the growth premium being assigned to AI-exposed names would compress.
For investors trying to think through their positioning, the practical question is not whether Nvidia will report good numbers tonight. It will. The practical question is whether the company’s outlook will validate the consensus assumption of continued strong growth through 2027 and into 2028, or whether the language will be incrementally more cautious in ways that the market will read as a tell. Hedge funds have voted with their books. They are not staying long into the print.
What Retail Investors Should Take From This
For Greg’s clients and any individual investor watching this play out, the hedge fund selling is a data point worth knowing but not necessarily an instruction to act. Professionals manage their books to a different risk framework than long-term retail investors do, and the urgency they feel ahead of a single earnings event is not the same urgency that should drive a 401(k) allocation decision. That said, the breadth narrowing, the cash-to-stocks rotation, and the prime brokerage flow data all point in the same direction, which is that the easy money in the current AI rally has likely been made.
A reasonable response for a long-term investor is to rebalance modestly, to take some of the largest gains in the most concentrated positions, and to redeploy into either cash or into the equal-weight components of the index that have lagged the mega-cap winners. That is a defensive posture, not a bearish one. It preserves the upside of continued AI strength while reducing the downside if the hedge fund community turns out to have read the setup correctly.
The other lesson is broader. When the prime brokerage data shows a record sale of single names in the week before an earnings event, the professional community is telling you something. You do not have to act on it. But you should know it is happening, and you should consider what it implies about your own positioning.
Frequently Asked Questions
How much did hedge funds sell last week?
According to Bank of America’s prime brokerage data, hedge funds sold $4.6 billion in single stocks last week, which the bank described as the largest amount sold outright in its records. Of that total, $3.01 billion was in technology stocks and exchange-traded funds, making the move heavily concentrated in the chip and AI complex.
Why does this happen right before Nvidia's earnings?
Hedge funds that want to be flat ahead of a binary event typically reduce their exposure in the trading week preceding the report. Nvidia’s earnings are the single most market-moving event in the current cycle because of the company’s index weight and its position at the center of the artificial intelligence supply chain. The selling pattern indicates professionals believed the risk-reward into the print was unfavorable.
What are analysts expecting from Nvidia's report?
Analysts polled by LSEG forecast that Nvidia’s bottom line more than doubled year over year, and that revenue grew by nearly 80 percent. Those are extraordinary figures by any standard, but they reflect the current consensus rather than the upside case. The market’s reaction will hinge on the company’s outlook and on what management says about the durability of AI capital expenditure.
Does this mean retail investors should sell tech stocks?
Not necessarily. Hedge fund positioning is one of several signals worth considering, but it operates on a different time horizon than most retail strategies. Long-term investors might consider rebalancing modestly, taking some gains in the most concentrated positions, and broadening exposure across the rest of the index. That preserves upside while reducing single-name risk.
What does Bank of America's "bull capitulation" signal mean?
Bank of America strategist Sarah Min has flagged the rapid rotation of cash into stocks as a contrarian sell signal. The framework holds that when investors deploy cash into equities at the current pace, there are few incremental buyers left to absorb supply at higher prices, which historically has coincided with near-term market tops. The hedge fund selling is consistent with that view.
Which sectors beyond chips could be affected?
The artificial intelligence trade has pulled in data center REITs, electrical utilities with AI exposure, copper miners feeding the cabling buildout, and AI-positioned software companies. A softening in Nvidia’s outlook would likely reprice each of these adjacent trades to varying degrees, with copper and the highest-multiple AI software names most exposed.
