Today, $NVIDIA(NVDA.US) is showing a strong trend, and $Apple Inc.(AAPL.US) is also performing well. Bros, is $Tesla(TSLA.US) coming to pick people up? I missed the rally! All in all, the night market is going strong tonight. Let's go!

🔥🎯 Nvidia's endorsement + hundreds of millions of dollars bet: $Lumentum(LITE.US) is betting on a bigger AI infrastructure gateway

My first reaction to this news wasn't "capacity expansion," but—
AI competition has started to shift from models down to the level of "light."

This move by Lumentum Holdings isn't just a simple factory expansion; it's a direct bet on the most critical yet often overlooked component in AI data centers: optical components.

And more crucially, NVIDIA has already taken sides in advance.

Conclusion first:
This isn't a "jumping on the AI bandwagon" move, but a clear strategic positioning in the supply chain.

When computing power demand explodes, the bottleneck gradually shifts from GPU performance to "how to transmit data at high speed," and the importance of optical communication will rise exponentially.

What Lumentum is doing here is securing this position in advance.

Let me break down the core logic of this deal.

First layer: Not building a factory, but "directly taking over mature capacity"

Lumentum acquired an already operational manufacturing facility from Qorvo.

What does this mean?

It's not a capital-intensive cycle from scratch, but rather:

Time compression
Risk reduction
Faster capacity ramp-up

Even the team comes "with the factory," which is a critical point.

Many people underestimate one thing:
Optical component manufacturing isn't something you can just do with money; it requires accumulated experience.

This move is essentially buying time.

Second layer: The technology path is locked in

The factory's future core products:

Indium Phosphide (InP) optical components
Continuous wave lasers
Ultra-high-power lasers

These aren't ordinary communication devices; they are "high-speed optical engines" specifically prepared for AI data centers.

Why is this important?

Because the bottleneck in AI training is shifting from "insufficient computing power" to "data can't be moved fast enough."

No matter how powerful the GPU is, if data transmission between nodes is slow, overall efficiency collapses.

Optical interconnect is the core technology to solve this problem.

Third layer: The most critical point—customers are already confirmed

Not "there might be demand," but:

Nvidia is already a customer.

The implication of this is very direct:

Capacity = has buyers
Investment = has a return path
Expansion = has certainty

This is also why the market is re-evaluating Lumentum's strategic position.

It's no longer just an optical communications company; it has entered the core chain of AI infrastructure.

Fourth layer: The timing is also subtle

Capacity ramp-up time: Mid-2028

Many might think that's too far away, but I think:

This is precisely the real pace of AI infrastructure development.

Now (2025-2026) is the GPU expansion cycle
Next (2026-2028) is the network and interconnect upgrade cycle

Lumentum is laying the groundwork in advance for the "next-stage bottleneck."

This is more important than chasing trends.

Looking deeper, behind this move is actually a bigger shift:

AI competition is turning from a "computing power race" into a "systems engineering race."

Whoever can control:

Computation (GPU)
Storage
Networking (optical interconnect)

Truly masters efficiency.

And Lumentum's step is equivalent to securing a seat in the "networking layer" in advance.

So what I'm more focused on now isn't:

Whether this investment will be profitable

But rather:

Will it become an underestimated link in AI infrastructure?

If the GPU is the "engine,"
then optical components are the "blood vessels."

No one ignores the engine, but what truly determines efficiency is often whether the blood vessels are clear.

Here's the question:

While the market is still fixated on GPUs,
will you start paying attention to these "invisible core components" in advance, or wait until they are repriced?