We Are Entering the Third Generation of AI

Dear Reader,

This is Dylan Jovine with Behind the Markets.

Happy Thursday. Today is Thursday, May 7th. What a great day it is to be alive.

We are entering the third generation of AI, and it is unfolding right before our very eyes.

The Cambrian Explosion

I keep thinking about the Cambrian explosion — that moment in the history of life on earth when different animals started evolving rapidly, each adapting to what nature selected for.

That is what is happening in AI right now.

Different models, different applications, different capabilities — all evolving fast, all competing for dominance.

The Three Generations

The first generation was AI that could answer questions.

You logged into ChatGPT, asked something, got a comprehensive answer.

Like a very smart search engine that didn't hallucinate too often.

The second generation was AI that could create — write, code, generate images, synthesize information.

The third generation — which is where we are right now — is agentic AI.

AI that does things for you.

Here is how we are using it at our company.

We have data coming in constantly — marketing numbers, customer acquisition data, financial analysis on companies we're following.

Instead of having someone spend their morning checking database after database and formatting a report, an AI agent does all of it and serves it up exactly the way you need it.

You want it medium well, they serve it medium well.

You want it medium rare, they serve it medium rare.

That is agentic AI.

What AMD's Earnings Told Us

You can see the third generation playing out in the earnings numbers.

AMD just reported blowout results — data center revenue up 57% to 60%.

We recommended AMD at $157.

It's now at $422.

A company of that size jumping 20% after a single earnings report is extraordinary.

But here is the number I keep coming back to.

The GPU to CPU ratio.

When ChatGPT first launched, for every eight Nvidia GPUs sold, one CPU was sold.

Now it's four GPUs to one CPU.

Intel's CEO expects it to reach parity — one CPU sold for every one GPU.

And Evercore has suggested it could flip entirely, with eight CPUs sold for every one GPU.

That shift is why semiconductor stocks have gone basically parabolic.

And it is not hype.

You can see it in the actual earnings numbers.

Where This Goes

AMD is already close to a half trillion dollar market cap.

I can see it entering the trillion dollar club over the next six to twelve months as it keeps posting results like these.

Intel is also sitting at around $500 billion now — up from the doghouse not long ago.

And ARM Holdings, the smallest of the three at around $200 billion, may be the best positioned of all to capture this shift.

I can see ARM going to a trillion dollars too as its results start to match the growth of its peers.

We have been in all of these names in our model portfolio since they were really, really cheap and nobody wanted them.

I am excited as an investor.

Because we have a generational opportunity here.

It’s not often that massive tech revolutions take place.

Historians will recognize a “before AI” era and an “after AI” era.

And the biggest opportunity I see right now is getting in on the stocks fueling this AI revolution.

Every one of these data centers, every GPU, every AI query processed — needs power.

More power than the grid was ever built to handle.

I've spent the last several months studying what I think is the single most important energy investment of the next decade.

One company. Sixty years of operating history. Google just signed a 15-year contract with them. The Pentagon declared their technology their number one energy resilience priority.

And on July 4th, the government hands them an 8-year monopoly.

Anyway, that's all I have for you today.

Have a wonderful Thursday.

I will speak to you tomorrow.

“The Buck Stops Here,”