On August 30, U.S. stocks fell across the board, especially technology stocks, and the Nasdaq fell more than 1%. Among them, Nvidia achieved the protagonist of "dragging its feet", and its market value shrank by $145.6 billion (about 1,038.2 billion yuan) overnight.

Although Nvidia's financial report released on Wednesday showed that revenue and profit exceeded market expectations, which should be good news, there are two "small flaws" that disappointed investors: first, the revenue of its most profitable data center business was slightly lower than expected; Second, the company's revenue forecast for the third quarter is only barely higher than analysts' estimates.

To put it simply, investors have too high expectations for Nvidia, even if it performs well, as long as it does not "explode", the market will still not buy it, and the stock price will fall.

Why is the market so sensitive to the $200 million gap?

Looking at the data alone, Nvidia's financial report can be said to be a "scholar-level" performance:

Quarterly revenue reached $46.74 billion, up 56% from the same period last year, slightly exceeding market expectations; The core data center business revenue rushed to $41.1 billion, a year-on-year increase of 56%, a new high; It earned $1.05 per share, which was also higher than expected.

It stands to reason that this should be a moment of "applause and celebration". But Wall Street doesn't seem to buy it - stock prices have plummeted. Why?

The problem is that "the details are not full marks". While data center revenue hit a new high, it fell short of analysts' expectations of $41.3 billion by $200 million. Just such a little "small deviation" was immediately interpreted by the market as: Have major customers begun to tighten their AI budgets?

Behind this, in fact, investors' expectations for Nvidia are too high. Baird analyst Ross Mayfield put it bluntly: "The market has become accustomed to Nvidia soaring all the way and now demands almost 'perfect' from it." Even a sneeze can be taken as a precursor to a cold. ”

Despite the fluctuations in stock prices, professional institutions still support it. Wedbush analyst Dan Ives commented on the financial report as "a bellwether for the AI industry" and emphasized: "There is currently only one chip company that is driving the global AI revolution, and that is Nvidia." He also predicted that Nvidia is expected to reach a market capitalization of $5 trillion in early 2026. At present, a number of investment banks have also raised their target prices.

To sum up: the short-term market is a bit "nervous", but in the long run, most people still believe that Nvidia is still the brightest star in the AI era.

Nvidia: Not a "member" of the US stock market, but the "heart" of the entire market

Recently, there is a saying that has become more and more widely circulated on Wall Street: "Now it is not Federal Reserve Chairman Powell who determines the rise and fall of US stocks, but Nvidia CEO Jensen Huang." "Sounds exaggerated? But the data tells you: it may be true.

This situation of "one family dominant" has a very far-reaching impact. Because many people now invest in index funds and ETFs (exchange-traded funds), these products will automatically "buy, buy, buy" according to the company's market capitalization. As a result, as long as money flows into these funds, Nvidia will continue to be bought.

What's even more exaggerated is that the data shows that the number of leveraged ETFs (that is, high-risk products with "leveraged" that are specifically betting on Nvidia) is actually as many as the ETFs that track the entire $52 trillion S&P 500 index! A large amount of money is tied to Nvidia's stock price, and when it shakes, the market has to shake.

The data is more intuitive: in the first half of this year, a full 35% of the market value of the S&P 500 index rose was contributed by Nvidia! If Nvidia's stock price suddenly falls 25%, the entire S&P 500 may fall 4.4% - this is "pulling the whole body".

JPMorgan also pointed out that individual investors accounted for 18% of U.S. stock trading volume, doubling from a decade ago. Most of them enter the market through index funds, thinking that they are buying diversified investments in "500 companies". But the reality is that the money in these 500 families is increasingly concentrated in the head, especially Nvidia. The so-called "diversification of risk" has become a "concentrated bet" to some extent.

Some people are worried that Nvidia is another "Cisco in the new era" and will repeat the dot-com bubble? But in terms of money-making power, it is much stronger than Cisco back then. Cisco's average profit margin between 1996 and 2000 was 17.2%, while Nvidia's average net profit margin in the past five years was as high as 40.34% - almost double.

To put it bluntly, Nvidia is not only "big", it is also particularly profitable. The market loves and fears it, relying on it to drive growth, but also worried that it is "too big to fail". This may be the trouble of "top companies": you have to win all the time, win to perfection, otherwise the market will turn around.

|Nvidia's "perfectionism" shackles

Starting from 2025, the contribution of investment in AI data centers to GDP has exceeded that of residents, which is the first time in history! Some analysts even say that without this wave of AI construction fever, the U.S. economy may have begun to shrink long ago. In other words, AI infrastructure may be helping the United States "delay the recession".

And all this is almost impossible to avoid Nvidia. Why? Because a large part of the AI money spent by global technology companies ends up going to Nvidia's pocket.

Nvidia CFO Colette Kress recently revealed that in the latest quarter, large cloud service providers (such as Microsoft, Amazon and other technology giants) alone contributed half of their data center revenue. The data center business itself accounts for 88% of NVIDIA's total revenue - it is an absolute "cash cow".

What's even more exaggerated is that Nvidia's future revenue is increasingly dependent on a few "big customers". At present, there are two mystery customers (Microsoft and Meta are generally considered in the industry) to account for 23% and 16% of its revenue respectively, adding up to almost 40%! This proportion has increased significantly from last year, indicating that Nvidia's "customer concentration" is getting higher and higher.

These giants are not relentless: Google's parent companies Alphabet, Microsoft, Meta and Amazon will invest a total of $400 billion this year, most of which will be used to build AI infrastructure. Nvidia predicts that by 2030, total global AI infrastructure spending will reach $3 trillion to $4 trillion - equivalent to the current level of total global military spending!

But here's the problem: Nvidia is too important. It is not only a company, but almost the "infrastructure" of the entire AI economy. Its growth continues, but so are the risks - too concentrated customers, too high market expectations, and the entire economy is staring at it. Once it "takes a breather" one day, the entire market may shake three times.

So, Nvidia is now like a giant standing on top of a mountain: the scenery is infinite, but the wind is also very strong. Everyone expects it to continue to grow "perfectly", but everyone knows that no one can never make mistakes.

Jingtai investment advice: embrace the trend, but stay awake

1. Long-term: Continue to hold AI core assets. Nvidia is still the "hydropower coal" of AI computing power, and as long as the global AI arms race continues, its moat will be difficult to shake. Long-term investors are advised to continue holding but avoid heavy positions in a single stock.

2. Medium-term: Pay attention to the spread opportunities of the "AI industry chain". Instead of just staring at NVIDIA, it is better to look at the entire "AI infrastructure chain": upstream: TSMC (foundry), ASML (lithography machine); Midstream: optical modules (such as Coherent), servers (such as ultra-micro computers); Downstream: AI applications (such as Microsoft Copilot, Google Gemini). These links also benefit, but with more reasonable valuations and more diversified risks.

3. Short-term: Be wary of "data-dependent callbacks". Whenever US economic data is weak, the market is worried about "whether AI spending can be sustained."

Recommendation: Pay attention to capital expenditure guidance (such as CAPEX in Microsoft, Meta's earnings reports); Pay attention to NVIDIA's supply chain scheduling data (such as GPU delivery date, price); Set a stop loss to prevent sharp fluctuations caused by "poor expectations".