To The Moon

Jensen Huang basically gave investors a roadmap for the AI buildout.

And the interesting part is that $NVIDIA(NVDA)$ isn't the whole story.

NVIDIA describes AI as a five-layer stack running from energy → chips → infrastructure → models → applications. Every layer has to be built, and every layer creates a different investment opportunity.

Here’s how I’m looking at it:

⚡ Layer 1 — Energy

$Bloom Energy Corp(BE)$ $GE Vernova Inc.(GEV)$ $Quanta(PWR)$ $Forgent Power Solutions, Inc.(FPS)$ $FuelCell(FCEL)$ $Vicor(VICR)$ $Vistra Energy Corp.(VST)$ $Navitas Semiconductor Corp(NVTS)$ $BWX Technologies Inc(BWXT)$ $Constellation Energy Corp(CEG)$ $Uranium(UEC)$ $Cameco(CCJ)$

Before you can run an AI factory, you need electricity.

And this may become one of the biggest bottlenecks in the entire AI cycle.

NVIDIA is already helping secure land, power and other infrastructure for AI factories, while data-center power demand is putting increasing pressure on grids.

So the AI trade isn't just about GPUs.

It's increasingly an electricity trade.

🔥 Layer 2 — Chips

$NVIDIA(NVDA)$ $Taiwan Semiconductor Manufacturing(TSM)$ $Intel(INTC)$ $Broadcom(AVGO)$ $Advanced Micro Devices(AMD)$ $ASML Holding NV(ASML)$ $ARM Holdings(ARM)$ $Marvell Technology(MRVL)$ $Applied Optoelectronics(AAOI)$ $AXT Inc(AXTI)$ $SanDisk Corp.(SNDK)$ $Micron Technology(MU)$ $SK hynix(SKHY)$ $Coherent(COHR)$ $Lumentum(LITE)$ $Aehr Test(AEHR)$ $Western Digital(WDC)$ $Corning(GLW)$

This is where most investors started.

GPUs, networking, memory, optical components, foundry capacity, lithography and advanced packaging all have to scale together.

That's why AI demand can spread far beyond $NVDA.

A faster GPU doesn't help much if you don't have the memory, networking, power delivery or manufacturing capacity around it.

🏗️ Layer 3 — Infrastructure

$NEBIUS(NBIS)$ $IREN Ltd(IREN)$ $CoreWeave, Inc.(CRWV)$ $APPLIED DIGITAL CORP(APLD)$ $Vertiv Holdings LLC(VRT)$ $Dell Technologies Inc.(DELL)$ $Hewlett Packard Enterprise(HPE)$ $Oracle(ORCL)$

This is where chips become usable compute.

Data centers. Servers. Cooling. Networking. Cloud capacity.

And this layer is becoming increasingly capital intensive.

NVIDIA has even partnered with major financial institutions to mobilize more than $500 billion of third-party capital for AI infrastructure buildout.

That tells you something important:

AI infrastructure is becoming an asset class of its own.

🧠 Layer 4 — Models

$Alphabet(GOOGL)$ $Meta Platforms, Inc.(META)$ $Amazon.com(AMZN)$ $SpaceX(SPCX)$

Now we move up the stack.

The models turn raw compute into intelligence.

But here's the catch:

The better the models become, the more compute they tend to consume.

So stronger models can actually reinforce demand for the first three layers.

💻 Layer 5 — Applications

$ServiceNow(NOW)$ $Palantir Technologies Inc.(PLTR)$ $DigitalOcean Holdings, Inc.(DOCN)$ $Snowflake(SNOW)$ $Cloudflare, Inc.(NET)$ $Fastly, Inc.(FSLY)$ $GitLab, Inc.(GTLB)$ $Tesla Motors(TSLA)$

This is ultimately where AI becomes a business.

Healthcare. Software. Cybersecurity. Robotics. Autonomous driving. Enterprise automation.

And this is also where the upside could become much larger — because applications can potentially capture far more economic value than the infrastructure underneath them.

That's why I don't think the AI investment thesis is simply:

“Buy $NVDA.”

It's a chain:

⚡ Energy powers AI
🧩 Chips create compute
🏗️ Infrastructure deploys it
🧠 Models turn compute into intelligence
💻 Applications monetize it

And the further down the stack you go, the more uncertain the winners become.

But the further up you go, the bigger the potential value capture can be.

That's the real opportunity.