U.S. software stocks are experiencing the most intense short-selling attack in over a decade. According to Morgan Stanley data, the daily short-selling volume in the software and SaaS sectors on Wednesday and Thursday reached one of the highest levels since 2010, following a brief short-covering rally that quickly faded. Goldman Sachs' trading desk described the market sentiment as "sell first, ask questions later," reflecting panic.

Hedge funds rapidly resumed short-selling strategies after a brief period of covering. A Morgan Stanley report indicated that the new short interest added from Wednesday to Thursday was only slightly below the historical peak recorded on January 29, with the software sector even exceeding the levels seen at the end of January. Infrastructure software stocks, which had previously seen significant short covering, are once again under heavy selling pressure.

Concerns about AI replacing human jobs are spreading to more sub-sectors. Market sentiment was impacted by comments from Microsoft's AI division CEO suggesting that "most white-collar jobs will be replaced by AI within 12 months." Transportation and logistics company C.H. Robinson plummeted by eight standard deviations amid fears that AI chatbots could reduce labor needs. The S&P 500 fell approximately 1.55% overnight, marking its second-largest decline in three months.

Defensive sectors are increasingly outperforming cyclical stocks. Goldman Sachs' cyclical versus defensive pair trade recorded its worst two-day performance since "Liberation Day," declining by more than 350 basis points cumulatively. The 10-year U.S. Treasury yield approached a three-month low near 4.08%, while the VIX index closed above the 20 level, indicating rising risk aversion.

Goldman Sachs' technology trading head Callahan noted that this is one of the most volatile trading environments he has ever seen. As the Russell Tech Index retreated to its 200-day moving average, many stocks in the technology and growth sectors showed oversold conditions, raising the question of which stocks may be excessively discounted.

Morgan Stanley: Historic Short-Selling Wave Returns In a report titled "Hedge Funds Resume Shorting, Software Stocks See Largest Single-Day Increase in Short Interest Since 2010," Morgan Stanley highlighted that hedge funds shifted to aggressive short-selling on Wednesday and Thursday, reversing the recent short-covering trend in U.S. equities. New short positions were highly concentrated in the software sector, ranking among the top five single-day increases since 2010.

Software and SaaS stocks became the focal point of short-selling, with new short interest surpassing levels seen at the end of January. Infrastructure software stocks, which had previously experienced heavy short covering, were not spared and are again facing significant selling pressure. In contrast, long position reductions were limited, meaning the overall net selling pressure remains lighter than during the late-January event.

Morgan Stanley data showed that last week saw the largest single-stock short-selling volume on record. Although short covering was intense on Friday and early this week, its duration was much shorter than expected, as the market quickly shifted from a "dead cat bounce" to a new wave of selling.

Goldman Sachs: A Market with "Nowhere to Hide" Goldman Sachs' trading desk wrote in its closing summary, "Today, there was nowhere to hide, with a 'sell first, ask questions later' sentiment pervading the market." Selling accelerated into the close, with no clear catalyst beyond the spread of AI concerns to more sub-sectors.

Some traders attributed the sell-off to a Financial Times report quoting Microsoft's AI division CEO as saying that "most white-collar jobs will be replaced by AI within 12 months" because "model programming capabilities have surpassed humans."

C.H. Robinson became another point of pain, with its stock falling by eight standard deviations. The market debated whether the company would become a loser due to AI chatbots matching freight more efficiently, potentially reducing labor needs.

Another Goldman Sachs trader previously noted, "The magnitude of volatility is severely dampening market sentiment. No one is willing to step in and buy the dip."

Index Decline, Tech Stocks Sold Off Collectively The S&P 500 fell approximately 1.55% overnight, marking its second-worst performance in the past three months. Goldman Sachs' cyclical versus defensive pair trade recorded its worst two-day performance since "Liberation Day," declining by more than 350 basis points cumulatively, signaling increasing risk aversion.

The 10-year U.S. Treasury yield fell to around 4.08%, near a three-month low, while defensive sectors outperformed. Verizon has risen in 16 of the past 18 trading sessions. The VIX index closed above 20 once again.

The market has become extremely sensitive and nervous about the potential compounding, disruptive, and spreading effects of AI. At the same time, large tech AI spenders and computing companies are also seeing valuation declines—Amazon has fallen for eight consecutive days, declining in 11 of the last 12 sessions; Google has dropped in seven of the last eight days; Nvidia and Broadcom have remained flat since last summer.

Defensive stocks continue to break out (Verizon, AT&T, Johnson & Johnson, Walmart, etc.), while the "AI infrastructure" theme is diverging (Compute vs. EMS vs. Memory). "Growth stocks" in software, internet, and payments are experiencing highly correlated sell-offs (SaaS, fintech, e-commerce, advertising, gaming, marketplace platforms, etc.).

Is the Valuation Adjustment Excessive? Goldman Sachs' technology trading head Callahan stated that this is one of the most turbulent trading environments he has ever seen. From his perspective, during this earnings season and market backdrop, stock price movements are more influenced by narratives around the AI ecosystem (such as large language model updates, blogs, and commentary snippets) than by earnings reports themselves.

The market is quickly pricing in perceived technological changes, but it is now worth discussing whether the market has moved "too far ahead" in some areas—meaning stock price changes are outpacing business developments. Callahan cited comments from two companies' earnings calls as examples (Tyler Technologies and Take-Two Interactive have not returned to pre-earnings levels):

Tyler Technologies stated during its earnings call: "There is a lot of market noise, but in the public sector, technology alone cannot win. For over 25 years, Tyler has guided clients through wave after wave of change, and our approach has remained consistent… Clients do not want additional tools that add complexity. They want practical AI that is deeply integrated into existing systems, properly governed, and solves real problems in a reliable and trustworthy manner."

Take-Two Interactive, when asked about Google's Genie, responded: "Frankly, I'm a bit confused. The video game industry has been built on machine learning and artificial intelligence since its inception. We use technology to create games in computers. Since the generative AI topic emerged 18 months ago, I have been enthusiastic about the future… This is only a small part of our work. If this product, this tool, materializes, it will make a part of our work better and more efficient."

Callahan noted that as the Russell Tech Index retreats to its 200-day moving average, many stocks in the technology and growth sectors appear oversold, making it worthwhile to discuss which stocks may be excessively discounted.

For reference, large-cap tech stocks (i.e., the Magnificent Seven) have underperformed the market by about 7.5 percentage points over the past few months, representing a relatively large pullback not driven by a "market event" (previous 10%-12% declines occurred during market events).

Callahan is watching whether the Magnificent Seven can stabilize, as this could serve as a potential anchor for tech stock stability (potential catalysts include Nvidia/Broadcom earnings, conference season, GTC大会, etc.).