Is the U.S. AI Stock Rally a Bubble? Watch Nvidia, Data Centers, and Sentiment

Every time AI stocks make new highs, someone calls it a bubble. That is not necessarily wrong, but it is also not very useful.

A bubble is not simply "prices are high." A real bubble usually has three things at once: fundamentals still exist, valuation runs too far ahead, and the crowd stops wanting to hear the bear case.

To judge whether the AI rally is still healthy, I would watch Nvidia, data-center capex, power constraints, valuation, and sentiment.

First: AI Is Not a Fake Theme

AI is not a theme with only white papers and no revenue. Cloud companies are buying chips. Data centers are being built. TSMC is taking advanced-node demand. Server and cooling suppliers are seeing real business.

So saying "AI is a bubble and everything goes to zero" is too lazy.

The better question is: the AI fundamentals may be real, but have stock prices pulled too much future growth into the present?

Nvidia Is the Thermometer, Not the Whole Universe

Nvidia matters because it is the central beneficiary of AI capex. Its revenue, margin, orders, and product timing influence the entire AI trade.

But markets sometimes treat Nvidia's strength as a guarantee for every AI-related company. That is dangerous.

Nvidia being strong does not mean every AI software company will make money.
Nvidia being strong does not mean every data-center supplier can keep high margins.
Nvidia being strong does not mean valuation can expand forever.

Core companies can be excellent while the outer theme becomes crowded.

Data-Center Capex Is the Second Key

For the AI rally to continue, cloud and mega-cap tech companies must keep spending real money. Not just talking about AI, but actually building infrastructure.

Watch:

Are cloud companies still raising data-center spending?
Is GPU and accelerator demand still strong?
Are power, cooling, land, and networking becoming bottlenecks?
Are management teams starting to talk more about discipline and efficiency?

If large customers slow spending, AI supply-chain stocks may react before the headlines fully catch up.

Power Is an Underestimated Risk

AI is not only a chip problem. It is also a power problem.

Data centers need electricity, cooling, transformers, land, networking, and water. Even if chips are available, infrastructure can slow the buildout.

This is not only negative. It can create opportunities in power equipment, cooling, and infrastructure. But for richly valued AI stocks, delays or cost increases can be amplified.

Valuation Is More Than a Quick PE Check

High-growth companies can deserve high valuations. The question is whether growth keeps up.

I would watch:

Is revenue growth slowing?
Have margins peaked?
Is customer concentration too high?
Is free cash flow keeping up with earnings?
Is the stock rising faster than earnings estimates?

If stock prices keep outrunning earnings revisions, the setup becomes fragile.

Sentiment: What Is the Crowd Saying?

I watch four social signals:

1. Research becomes belief

When risk discussion disappears and every question is dismissed as "you don't understand AI," that is a warning.

2. Core winners become name-only winners

When anything with AI in the name rallies, capital is searching for the next story.

3. Financials become screenshots

When discussion shifts from revenue, margins, and capex to profit screenshots, speculation is rising.

4. Pullbacks become impossible

If the crowd believes every dip is automatically a gift and no bad news can matter, expectations are too one-sided.

My Takeaway

The U.S. AI rally is not fake, but it is not risk-free.

A healthy AI bull market looks like this: core earnings keep moving higher, data-center spending continues, margins hold, pullbacks find buyers, and investors still discuss risk.

A dangerous AI bull market looks like this: prices rely mainly on narrative, weaker theme names rally, questions are punished, and valuation expands faster than earnings.

You can be bullish on AI. Just do not confuse a long-term trend with permission to buy any price.

FAQ

Q: Are AI stocks in a bubble?

A: Not as a single group. Core AI companies have real revenue and cash flow, but some theme names may already price in too much.

Q: Does Nvidia still represent the AI trade?

A: It is one of the most important thermometers, but it is not a guarantee for every AI stock.

Q: How can regular investors avoid chasing?

A: Do not look only at new highs. Watch revenue, margins, capex, valuation, and sentiment. Know what would prove your thesis wrong before buying.

閱讀中文版本： 繁體中文版本