AI News

Nvidia's Jensen Huang Steps Down; US AI Capital Expenditure May Enter a Slowdown Channel

Phoenix TechnologyUpdated Nov 16 20:34

AI summary

Nvidia CEO Jensen Huang has stepped down, with the CFO taking over and halting guaranteed financing for data center projects, seen as a signal that US AI capital expenditure is shifting toward contraction. The article states that the free cash flow of several major cloud vendors has turned negative, with off-balance-sheet hidden debt of about $4 trillion, while annual capital expenditure of $1 trillion requires corresponding annual revenue of $10 trillion. The author also mentions that Chinese open-source large models such as DeepSeek have blocked the possibility of US large models reaping windfall profits through monopoly by offering ultra-low prices.

Body · AI translation

The US AI capital expenditure is clearly about to enter a slowdown channel. Jindian Download Client Exclusive First Look November 16, 2026 07:48:00 From Beijing Updated November 16, 2026 07:59:17

This article is written based on public information and is for information exchange only; it does not constitute any investment advice.

Author: Yu Laotou

Jensen Huang's resignation from Nvidia is like blind men touching an elephant—the explanations offered are all over the place: some say it was discrimination against little yellow people, some say it was collateral damage from the two-party struggle in the midterm elections, and some even say it was because large US companies value the moral bottom line of their CEOs.

Of course, if you say discrimination, some people will disagree; the specific judgment ultimately depends on each person's standpoint.

01

When it comes to this wave of AI frenzy in the United States, Jensen Huang really was the leader. Nvidia is the central bank of AI infrastructure. Without Jensen Huang, the AI operation would never have expanded so large, cloud vendors' capital expenditure would never have been pulled to the trillion-dollar scale, and AI competition would never have been pushed to the level of national strategy.

In ancient Chinese history, what was valued was that a great man stands between heaven and earth and cannot remain depressed and subordinate to others for long. If one cannot leave a good name for a hundred generations, one should at least leave a bad name for ten thousand years. Regardless of how later generations judge him, Jensen Huang at least deserves a mention in the history books.

Relative to the resource endowments of the United States, the AI industry has a natural defect: hardware is too weak, actual implementation is too difficult, and it is easy to make wedding dresses for others. AI is an industry combining software and hardware; in the early stage it can rely on software, but later implementation inevitably depends on hardware.

In terms of hardware, the only area where the United States is strong is chips, and only in chip design. In foundry manufacturing, there is basically no domestic capacity at all; it all relies on Jensen Huang's silver tongue to mobilize an entire hardware supply chain in East Asia.

As for software, it appears very strong. But basic education in the United States has enjoyed peace for too long, declining from top students to mediocre students and now to poor students; high-tech work such as large models can only rely on little yellow people.

This has produced the following result: on the frontier track of AI, it is mainland Chinese competing against American Chinese.

This leads to a problem: AI data centers, these huge power consumers, can only be built near existing residential areas and piggyback on existing infrastructure. The result is that electricity prices for residents near data centers rise sharply, but jobs barely increase—the benefits go to you, and I bear the drawbacks.

This is also why many places in the United States and Europe have recently seen opposition to data center construction.

Under normal business logic, with $1 trillion in capital expenditure per year, under the premise of a reasonable rate of return, annual revenue of $10 trillion per year would be needed in the future. The problem is where this money will come from.

In 2025, global GDP will be $120 trillion, the food and beverage industry $8 trillion, and healthcare $11 trillion. Where will AI find this $10 trillion in revenue?

Moreover, with the current level of capital expenditure, Wall Street can no longer bear it:

The engine of capital expenditure—the free cash flow of several major cloud service providers—has turned negative, leaving no spare funds to buy back shares and prop up stock prices;

Off-balance-sheet hidden debt is about $4 trillion, equivalent to the hole left by at least ten Evergrandes;

Massive financing by companies in the large-model industry chain has pushed up financing rates for U.S. Treasuries and other enterprises;

High capital expenditure has pushed up prices of upstream chips and memory, causing consumer electronics demand to shrink;

The key point is that Chinese open-source large models represented by DeepSeek, at ultra-low prices, have completely blocked the possibility of U.S. large models obtaining windfall profits through monopoly.

So the next step is inevitably to increase revenue and cut expenditure.

On the expenditure-cutting side, Jensen Huang steps down, a CFO takes over, no longer providing guaranteed financing for various data center projects, and the AI central bank shrinks its balance sheet; large models de-Nvidiaize, or reduce the use of Nvidia chips, increase competition in the chip industry, and lower chip prices; slow down or halt data center construction.

On the revenue-increasing side, first is to remove the firewood from under the cauldron, block sales of Nvidia chips to the East, and dig a pit for the East's large models; second is to promote the expansion of downstream applications of large models, striving to achieve a breakthrough in the direction of embodied intelligence.

03

America's attempt to lead China in technology is over

Regarding Jensen Huang's departure, conspiracy theories on the internet are surging. Right or wrong is actually not important; what matters is what these currents of thought reflect.

For the AI industry, especially in large models, ethnic Chinese are certainly the backbone, and Jensen Huang's departure clearly plants the seeds of distrust for ethnic Chinese. Even if you have reached Jensen Huang's level—immigrated at age 9, received an American education, married a white woman—people who come later will also weigh whether they can reach Jensen Huang's level.

Top elites, whether in the past or the future, in the era of globalization, are inevitably priced globally.

More than two thousand years ago, Li Si's "Memorial on the Expulsion of Guest Ministers" already stated the consequences of this approach: "Now you abandon the common people to aid the enemy states, and reject guest ministers to help the feudal lords, causing the world's scholars to retreat and not dare to face west, and to halt their steps and not enter Qin. This is what is called 'lending weapons to bandits and supplying grain to thieves.'"

Of course, the ebbing of AI does not mean U.S. stocks will fall. Everyone also knows that the rise of U.S. stocks is not the result of the Federal Reserve and U.S. policy, but the goal of overall U.S. policy. U.S. stocks will still rise overall, but investing in U.S. stocks does not necessarily make money, because there is also the exchange rate in between.

For a track where originally everything rose together, after the style of the picture suddenly changed, I can only wish everyone good luck.

"Special Statement: The content above (including videos, pictures, or audio, if any) is uploaded and published by users of Dafeng Hao, a self-media platform under Phoenix Net. This platform only provides information storage space services.

Notice: The content above (including the videos, pictures and audios if any) is uploaded and posted by the user of Dafeng Hao, which is a social media platform and merely provides information storage space services."

Source

Phoenix Technology