Hey friend,

When I say “AI”, you probably think about ChatGPT or some other LLM.

Most people do. But those chatbots are just the tip of the iceberg.

There’s a massive tech stack sitting underneath them.

And as an investor, that’s important because depending on which layer you invest in, your money will behave very differently.

Investing in some layers will be considered high-risk, and I expect thousands of businesses to fail.

Other layers will be full of safer business models. Those companies will quietly scale into huge businesses beneath the risky ones making news headlines.

I watched Nvidia’s CEO, Jensen Huang, break these layers down last week.

But he broke them down for users and business analysts to understand.

Allow me to explain what it all means for you as an investor!

That’s a great way to explain the stack, so let’s start at the bottom.

You wouldn’t believe the amount of energy that AI needs.

In 2019, OpenAI consumed enough energy to supply 1,462 apartments with electricity for a month.

What’d they do with the energy?

They were just training ChatGPT-3.

Energy touches every layer in the AI stack: data centers need power. Chip factories need power. Cloud providers need power. AI applications need power.

Everyone eventually sends money down to the energy layer.

So, it’s ironic that I rarely hear anyone talking about it.

It’s one of the safest layers for your investments—which is why I mentioned IDACORP last week.

Not a single AI model has ever been built without semiconductor chips.

And that is why companies like Nvidia are so successful.

From a business perspective, it doesn’t matter which product wins in an AI-first world.

Whether it’s ChatGPT, Claude, Gemini, Groq, or some up-and-coming startup, Nvidia sells chips to all of them.

Currently, Nvidia dominates this layer by holding an estimated 80% to 90% of the competitive market.

The cloud is the last infrastructure layer needed to build and deploy an AI model.

Why?

Because building a private data center with thousands of AI chips would cost companies tens of millions of dollars upfront. And most lean companies don’t have that in their budgets.

By renting the same infrastructure through the cloud, they pay only a few dollars per hour for power.

This allows small startups to compete with tech giants without the ludicrous price tag.

And every quarter, Cloud providers make tens of billions selling these services to SaaS and PaaS companies, development agencies, indie developers, and others.

Historically, the safest and most profitable way to invest in this layer is to acquire stocks in the large- and mega-cap companies:

Nearly every web application you’ve ever used has gone through one of these companies.

Which means nearly every cloud-based AI application to ever be built will pay them.

This is where most people believe AI lives.

But as we’ve just discussed, all three infrastructure layers need to exist before an AI model can.

Nonetheless, many of the companies building today’s mainstream models are preparing for IPOs (Initial Public Offerings). These include:

If you’re interested in investing in these companies, here are their IPO timelines:

I suspect many more companies will appear in this layer. Many of which may fail.

To be successful at this layer, review the partnerships made with each AI model company.

If you’re looking for safety, Government contracts and defense partnerships will likely be at the top of your list.

This is happening right now.

This is where Jensen believes the most economic value will be created. And he's probably right.

Nearly every application I use is AI-powered or has an AI assistant. ChatGPT, Grammarly, Cursor, Notion, Microsoft Office. It’s everywhere.

Millions of new businesses will spring up, and millions of consumers will pay for AI-powered software.

But it’s also where hackers play. It’s where competition is the most dense. It’s where anyone can launch a business, and it has the highest risk of business failure.

It has always been that way.

When investing in this layer, I recommend being extremely cautious.

It’s worth noting that most application companies worth investing in are likely to be acquired by a large-cap company like Meta, Amazon, or Google.

…just food for thought.

Before I go, I want to leave you with something actionable.