Dear Reader,
This is Dylan Jovine with Behind the Markets.
Happy Thursday.
Today is Thursday, August 13th.
If there's one thing I want you to take away from this channel after all these years, it's this.
Return on investment.
Every decision — whether you're comparing stocks, evaluating a bond, or running your own business — comes down to the same question.
What is the best risk-adjusted return on my money?
And right now, the biggest question in markets is whether the most enormous CapEx spending spree in human history is going to generate a real return.
Let me walk you through the numbers.
The Scale of What's Being Spent
Hyperscalers have committed a record $2.6 trillion in future spending across data center leases and equipment purchases.
And here's something most people miss.
A lot of these obligations don't even appear on the balance sheets.
This is why I always tell the young people who work with me: amateurs study income statements, professionals study the footnotes.
Here's what the footnotes actually show.
Alphabet has $811 billion in purchase commitments and $85 billion in leases not yet commenced — nearly $900 billion in combined obligations.
Meta: $349 billion, with $279 billion showing up on the balance sheet.
Microsoft: $228 billion in purchases, $329 billion in leases not yet commenced.
Amazon: $130 billion across both categories.
Oracle: $32 billion in purchases, but $260 billion in leases not yet commenced — the largest lease obligation in the group.
That is an almost incomprehensible amount of money.
Why Are They Spending It?
Bloomberg just put out a forecast showing the market they're chasing.
Generative AI is expected to be a $2.3 trillion per year market by 2032.
It is expected to account for 22% of all tech spending across hardware, software, and services.
So here's how the math looks from their perspective.
They're committing $2.3 trillion in capital spending to access what they expect to be a $2.3 trillion per year market.
Over ten years at that rate, that's $23 trillion in revenue.
Investing $2.3 trillion to make $23 trillion.
That's the bet.
Who's Winning and Who Isn't
We've already seen which companies are generating a real return on this investment.
The companies with corporate cloud customers — Microsoft, Alphabet, Amazon — are showing the beef.
Their customers are clamoring for AI tools.
The infrastructure spending is meeting real demand.
The stock market is rewarding them for it.
The companies struggling are the ones that haven't translated the spending into actual revenue yet.
Oracle is spending enormously but not yet showing the return.
Meta is the one I keep coming back to.
Phenomenal underlying business.
But they're a consumer-facing company spending almost as much as the biggest cloud players — without yet figuring out how to monetize it at the same scale.
That's why the stock has been trading the way it has.
I still like it long term.
But the market is asking a fair question.
The Shift Nobody Has Fully Priced Yet
Here's the part of this story that I think Wall Street still hasn't caught up to.
The AI market has hit an inflection point.
Inference — not training — is becoming the dominant revenue driver.
Let me explain the difference.
AI training is the heavy learning phase.
The model is ingesting massive amounts of data and adjusting its internal parameters.
Think of it as the education phase.
AI inference is what happens after.
The model jumps out of the nest and starts to fly.
It takes everything it learned and applies it to real world decisions, in real time.
That's robotaxis.
That's Optimus robots.
That's automated systems making split-second decisions in the field.
That is what Elon Musk is betting his entire career on.
Not training.
Inference.
And the chips, the CPUs, the infrastructure that powers inference are very different from what powers training.
We've done a lot of research on this.
A lot of the positions we hold are focused specifically on the inference side of the equation — because that's where the real-world economic value gets extracted.
And Wall Street is only beginning to understand it.
We are in the second or third inning of this development cycle.
Maybe the fourth.
The opportunities in front of us are extraordinary.
And I bring you a few of my best ones here.
Have a wonderful day.
I'll see you tomorrow.
“The Buck Stops Here,”
Found this helpful? Share it with others.
Written by Dylan Jovine