[你懂的]  The market’s biggest story today isn’t simply whether AI stocks can keep rising. The more important question is: Where does the next wave of AI capital spending go?

The latest U.S. jobs data showed a sharp slowdown in employment growth, reducing expectations for another near-term Fed rate hike. That is supportive for growth stocks. But with long-term Treasury yields still elevated, investors have less room to pay any price for future growth.

That makes earnings power and real AI demand increasingly important.

And this is where I think the next opportunity may be hiding.

1️⃣ NVIDIA — Still the AI Core

NVIDIA remains the undisputed center of AI computing. Its massive buyback authorization also highlights the extraordinary cash generation of the business.

But there is a problem for investors:

Everyone already knows NVIDIA is great.

The bigger opportunity may be in the companies supplying everything required to make those GPUs actually work.

2️⃣ Micron — AI Memory Is Becoming a Bottleneck

AI servers don't just need GPUs. They need enormous amounts of high-performance memory.

Micron is increasingly becoming one of the key beneficiaries of the HBM and data-center memory boom.

The interesting part is that this may be more than another traditional memory cycle.

As AI clusters become larger, memory requirements rise with them.

More GPUs → more HBM → more memory demand → tighter supply → stronger pricing power.

That is the thesis worth watching.

3️⃣ SanDisk — The AI Storage Layer

AI models generate enormous amounts of data.

Training data, model checkpoints, inference workloads and enterprise AI applications all require storage.

That puts SanDisk in an interesting position.

The market spent years focusing on the GPU shortage. The next bottleneck could increasingly move down the stack:

Compute → Memory → Storage.

SNDK is therefore one of the names I would keep on the radar, especially on meaningful pullbacks rather than chasing vertical moves.

4️⃣ Credo — The “Invisible” AI Infrastructure

This is perhaps the most interesting part of the trade.

As AI clusters scale from thousands of GPUs toward much larger systems, moving data between chips becomes increasingly important.

That creates demand for high-speed connectivity and optical/electrical interconnect solutions.

This is where Credo Technology (CRDO) becomes interesting.

It doesn't sell the headline AI chip.

It helps connect the chips.

And sometimes the less glamorous part of the infrastructure can have surprising pricing power.

5️⃣ STX & WDC — A Potential Contrarian Setup

Seagate and Western Digital recently suffered sharp selling after reports that Toshiba plans to expand HDD production for AI data centers.

At first glance, this looks bearish:

More supply = lower pricing power.

But there is another way to look at it.

Why is Toshiba expanding HDD capacity in the first place?

Because AI data centers are demanding enormous amounts of storage.

The key question isn't whether AI needs HDDs.

It clearly does.

The real question is:

Will future supply growth exceed future AI storage demand?

If supply expands faster than demand, the bearish thesis wins.

But if AI storage demand grows even faster, the recent selloff could eventually look excessive.

That's why STX and WDC are now interesting to watch.

6️⃣ Bloom Energy — The AI Power Bottleneck

And finally, there is a completely different bottleneck:

Electricity.

AI data centers are becoming so power-intensive that access to reliable electricity could become as important as access to GPUs.

This creates opportunities in distributed generation, fuel cells and data-center power infrastructure.

Bloom Energy is one speculative name worth monitoring.

---

My AI Infrastructure Watchlist

🥇 MU — AI Memory

🥈 SNDK — AI Storage

🥉 CRDO — AI Connectivity

STX / WDC — Data-center HDD

BE — AI Power

NVDA — AI Computing

The bigger picture is what matters.

The AI trade may be evolving from:

GPU → Memory → Storage → Networking → Power

The first wave made NVIDIA famous.

The second wave may create entirely different winners.

And that's where I think investors should start looking.

The question isn't just “Who makes the AI chip?”

The better question is:

«“What does the AI ecosystem need to keep those chips running?”»

That could be where the next big opportunity is hiding.[思考]