Indiscriminate exit, what is the market really afraid of?
Author: Godzilla, Gelong
The global market during this period has indeed been distressing.
In the past, when the market adjusted, if the stock market fell, one would turn to gold; if equity was unstable, bonds would be the choice. When risk appetite shrank, there were at least some safe havens. However, this time is very different; global stock markets, bond markets, futures, gold and silver, and even cryptocurrencies are all being indiscriminately hammered down together.
And the extent of the decline is significant.
Since the fourth quarter, silver has dropped by 33%, gold by 16%, crude oil by 13%, copper by 15%, tin by 23%, and Bitcoin by 27%...
As for the stock market, let's not even mention the domestic situation; it's equally grim abroad.
In the past month, the Nasdaq has fallen over 15%, the S&P 500 over 8%, Japanese stocks over 20%, and Korean stocks over 30%.
Especially in this past week, the decline has been even more severe and rapid; historically, such a significant drop across all categories in the global market is rare.
Only bonds and cash have not fallen; instead, safe-haven funds are pushing up U.S. Treasury prices, with the yield on the 10-year U.S. Treasury falling back to around 5%, and the U.S. dollar index regaining 103.8.
In other words, apart from cash assets, all other assets are declining.
What exactly are people afraid of?
01
Historically, when major asset classes experience a simultaneous significant drop, there are usually two underlying drivers.
The first is a liquidity crisis, where the market genuinely lacks money, and institutions are forced to liquidate all sellable assets to meet margin calls and handle redemptions, leading to a stampede out of the market, which can easily evolve into a breakdown of the financial chain.
The second is a reassessment of interest rates, where the cost of money becomes more expensive, discount rates rise, and all long-term cash flow assets need to be revalued. Non-yielding assets like gold are the first to be affected, and high-leverage growth stocks that rely on financing for expansion will also be repriced.
But it seems that both driving factors are occurring simultaneously now.
This is a very rare situation in history.
Other markets are difficult to elaborate on, so let's focus on the U.S. stock market.
In this round, the hardest-hit sectors in the U.S. stock market are upstream and high-leverage segments. Let's rewind to October 6, when the Nasdaq had just reached a historical high, with the Philadelphia Semiconductor Index closing at 13,217 points, and Nvidia touching $243, with a market cap of about $5.87 trillion.
In the following month and a half, the Philadelphia Semiconductor Index fell to 9,419 points by November 19, a decline of about 29%; Nvidia dropped about 40% from its peak, with its market cap shrinking to around $3.52 trillion.
As for Google, it has retraced over 20% so far, which is not insignificant.
Looking at it layer by layer, the further upstream in the industrial chain and the higher the capital leverage, the heavier the fall. The new cloud vendor CoreWeave dropped over 40% a month before its earnings report, Nebius fell about 15% in a single day, data center developer Hut8 and cooling equipment manufacturer Modine each fell over 20% in one day, and Vertiv dropped nearly 13%;
In storage and optical communications, SanDisk fell 22% in a single day, down over 60% from its peak, Micron and Western Digital each fell over 17% in one day, Lumentum dropped over 18%, and Marvell fell over 16%.
Among the established giants, Intel dropped 45% throughout October, with a single-day plunge of 29% on October 30, marking the largest single-day drop since the 1970s, and has cumulatively fallen 54% since the fourth quarter began.
In the past two years, the logic for pricing AI companies in the U.S. stock market has been simple: the more aggressively capital expenditures increase, the higher the stock price rises, with the market only asking, "Is money being invested?"
However, during the earnings season at the end of October, this logic suddenly broke down.
Google Cloud's revenue grew by 103% year-on-year, yet the next day its stock price plummeted by 12.6%, evaporating about $500 billion in a single day; AWS's growth rate was 39.3%, reaching a nearly five-year high, yet it still dropped 7% after hours;
Meta's revenue increased by 26%, but its stock price fell by 7%. The act of spending money has shifted from being a positive factor to a negative one.
The exam has changed. Previously, it was about "who dares to spend money," but now it is about "can the money be earned back?"
Can AI investments be converted into revenue, profit margins, and free cash flow? Can data center utilization rates, service prices, and renewal rates hold up? These mid-term questions, which were previously overshadowed by high growth, are now all on the table.
I do not fully agree with the vague statement that "the AI narrative is fading," but while the narrative has not disappeared, the market is no longer willing to pay such high prices for imagined returns.
02
Currently, there are several layers of pressure on the U.S. stock market.
The first layer is cash flow and leverage, relating to the liquidity crisis mentioned earlier.
In the first nine months of this year, Microsoft, Google, Amazon, Meta, and Oracle's total capital expenditures amounted to approximately $556.9 billion, exceeding the operating cash flow of $542.9 billion during the same period; four of the seven giants have turned negative in free cash flow this year for the first time in a decade.
Google's Q3 free cash flow was -$9.07 billion, Amazon's free cash flow over the past twelve months was -$42.36 billion, with buybacks cut by 83% year-on-year, and the shortfall was covered by issuing bonds and increasing shares. The seven giants collectively shifted from a net buyback of $144 billion a year ago to a net issuance of $276 billion.
The high-leverage business model of new cloud companies is built on the premise that "financing can always be obtained cheaply." With interest rates remaining high, this batch of assets is the first to be repriced.
The second layer is policy.
After the midterm elections, the Democratic Party secured both houses, and new governors in key data center states like Texas, Georgia, and Ohio have indicated tighter approval processes. About 40% of the capacity waiting to be built nationwide falls in these states; voters are increasingly frustrated with data centers consuming water and electricity, raising living costs, and this frustration is turning into policy resistance.
The tech crash in U.S. stocks on November 5 was just the beginning of market concerns and is far from over.
The third layer is targeted attacks on tech giants.
So far, there have been continuous black swan events among U.S. tech giants.
Nvidia's leadership change is still unsettled, and market concerns have not been alleviated;
Last night, short-selling institutions began targeting AMD, bringing the "circular trading" allegations to the forefront, causing AMD to drop over 12% at one point.
The fourth layer is the potential risks of market sentiment and crowded positions. Previously, the storage basket had risen over four times in a year; once a reversal occurs, leveraged positions and profit-taking will concentrate on closing, leading to a faster decline than changes in fundamentals.
Of course, the fifth layer, which is also the most critical underlying factor, is the Federal Reserve's interest rate expectations.
This week, just before and after the October interest rate meeting, there was a flurry of hawkish comments from Federal Reserve officials, including several previously dovish officials, who clearly hinted at a rate hike in December. This was a major trigger for the sharp decline in U.S. stocks, which then affected the global market.
This creates a very strange situation.
All of the above factors are significant negatives, and their combined pressure is simply unbearable. This is precisely why the current market funds are most concerned.
03
However, there are several things that cannot be ignored.
I do not believe it is time to declare a "collapse of the AI narrative." Several counter-facts deserve to be presented.
First, the demand-side data has not deteriorated. Morgan Stanley estimates that the five major cloud vendors will have capital expenditures of about $1.125 trillion by 2027, up from $792 billion in 2026; Gartner predicts that global AI spending will reach $4.05 trillion in 2027, a year-on-year increase of 47.5%.
CoreWeave has a backlog of orders worth $132.5 billion, and Nvidia's GPUs are sold out, with a full order book. The market is questioning whether investments can be recouped, not whether orders have disappeared.
Second, valuations have already fallen in advance.
Nvidia's forward price-to-earnings ratio has returned to 11 to 12 times, a fifteen-year low, cheaper than the S&P's 18 times; although most institutions have lowered target prices, they still maintain buy ratings, with the consensus being "the business is not broken, just the multiples need to be reset."
Apple is the least spender on AI infrastructure among the seven giants, yet it has risen about 27% this year and is the most resilient. This indicates that the market is not blindly abandoning technology but is distinguishing between those with strong cash flow quality and those relying on leverage for growth.
Third, this round is more about giving back excessive gains rather than a full liquidation.
UBS's baseline scenario for 2027 still sees the S&P at 8,200 points; only in the pessimistic scenario of "AI investment stalling and computing power construction exceeding real demand" would we see 5,500 points.
These are two scenarios, not facts that have already occurred.
The next one or two quarters will be a verification period. Rather than guessing the direction, it is better to focus on several scoreable nodes:
Can Nvidia's next quarter guidance hold above $120 billion, and will its gross margin stop declining?
Can Alphabet's Google Cloud revenue reach $33 billion in early February 2027, maintain an operating profit margin above 30%, and will capital expenditures be cut?
Can Micron and other storage chain gross margins hold above 80%?
And then there’s the progress of the policy on the electricity bill and the financing environment for new cloud customers.
If most of these numbers are realized, there is still confidence in restoring the fundamentals.
04 Conclusion
If we look at the global and U.S. stock markets together, my judgment may be more restrained; we are currently at the end of a high valuation phase, transitioning into a phase of calm verification.
On a global level, money is indeed becoming more expensive, and assets are collectively being revalued. Short-term risks are prevailing, and sentiment is fragile; even a slight negative can trigger a significant drop; on the U.S. stock market AI front, demand has not yet been disproven by earnings reports, and valuations have squeezed out a lot of bubbles. The evidence for concluding that "the narrative has collapsed" is insufficient.
The current situation is indeed very unfavorable, and the upcoming period will likely continue to test.
However, if one sees a sharp drop and shouts that the bubble has burst, completely denying the AI narrative, that is not the case.
In the chaotic sell-off, quality assets and fragile bubbles are often pushed down together.
Rather than being driven to make decisions by the market's violent fluctuations, it is better to use this round of adjustments to calm down and conduct research and sorting, to see which assets are gradually being mispriced, and then patiently wait for the market turning point to provide a re-entry opportunity.