🔥🚨 The whole world is investing in these areas, and capital is flowing into these 7 directions. What I'm more concerned about is "which are real demands and which are just narratives."

On the surface, this looks like a very scattered list.

But when you break down these sectors, you can actually see a very clear structure—

It's not about industry dispersion, but the same logic unfolding across different layers.

The first layer is the ability to "leave Earth":

$Rocket Lab(RKLB.US)

$AST SpaceMobile(ASTS.US)

$Planet Labs(PL.US)

$Serve Robotics(SERV.US)

Space infrastructure is no longer science fiction, but part of communication, data, and even future networks.

The core of this line is not launch, but "who controls orbital resources and data entry points."

The second layer is the issue of "light":

$Applied Optoelectronics(AAOI.US)

$Lumentum(LITE.US)

$Coherent Corp.(COHR.US)

$Ciena(CIEN.US)

The bottleneck for AI computing power is shifting from computation to transmission.

Whoever can move data more efficiently holds the key to the next stage of performance improvement.

The third layer is "unmanned execution":

$Ondas(ONDS.US)

$Unusual Machines(UMAC.US)

$Aerovironment(AVAV.US)

$Kratos Defense & Security(KTOS.US)

Drones are not just military or consumer products; they are becoming a low-cost execution layer.

When decisions are made by AI, execution must be automated.

The fourth layer is "stable energy":

$Oklo(OKLO.US)

$NuScale Power(SMR.US)

$GE Vernova(GEV.US)

$Centrus Energy(LEU.US)

AI, electrification, and computing centers are essentially competing for the same thing—electricity.

And nuclear energy is currently one of the few options that can meet both scale and stability requirements.

The fifth layer is "computing power assetization":

$IREN(IREN.US)

$Nebius(NBIS.US)

$Cipher Digital(CIFR.US)

$Coreweave(CRWV.US)

The core here is no longer "how strong is the computing power," but "how is the computing power sold."

Electricity + GPUs are becoming a new category of infrastructure assets.

The sixth layer is the "inference layer":

$CloudFlare(NET.US)

$Digitalocean(DOCN.US)

$Fastly(FSLY.US)

$One Stop(OSS.US)

Training is just the beginning; the real scale is in inference.

Whoever can deliver AI services stably and at low cost is closer to a commercial closed loop.

The seventh layer is "electricity itself":

$Vistra(VST.US)

$Constellation Energy(CEG.US)

$Vertiv(VRT.US)

$Bloom Energy(BE.US)

Ultimately, all the stories above will fall on one constraint—whether there is enough electricity.

So these 7 sectors are not actually 7 opportunities, but a chain:

Energy → Computing Power → Transmission → Inference → Execution → Application → Expansion (Space)

But the real key is not "where the money is flowing," but:

Which demands have already occurred,

And which are just market imaginations priced in advance.

If you only look at the narrative, all sectors are valid;

But if you look at the speed of implementation, the differentiation will be very obvious.

I tend to prioritize looking at three things:

Whether electricity truly becomes a bottleneck,

Whether inference demand continues to amplify,

Whether computing power can be stably monetized.

These three points determine whether this chain can be closed.

So the question is not whether to participate in these sectors,

But rather—are you buying "future certainty" or are you buying "narrative priced in advance"?

If you had to choose one direction to prioritize betting on, do you value electricity, computing power, or the inference layer more?