$700 Billion. Then $1.5 Trillion. The AI Infrastructure Bet Is Getting Harder to Ignore.

🎙 THE CLIP — HERO QUOTE

“So we’re looking this year, it’s looking like big tech alone will put something like $700 billion in towards CapEx this year. I think last year was something like $350 to $400 billion. And next year is projected to be $1.5 trillion in buildings.”

— Alex Kantrowitz (@kantrowitz)
Clip via Alex Kantrowitz (@kantrowitz) on Big Technology Podcast with Paul Kedrosky / Big Technology Podcast — Why The AI Bubble Will Burst: The Most Logical Case — With Paul Kedrosky

THE SCALE — AS KANTROWITZ FRAMES IT

Big tech CapEx, per his argument: ~$350–400B last year → ~$700B this year → $1.5T projected next year in buildings alone. Each step is not incremental. It is exponential commitment to a thesis that has not yet fully proved out at the revenue layer.

📐 THE STRUCTURAL READ — FOURWEEKMBA ANALYSIS

This is where the FDE Framework sharpens the picture. The companies driving this CapEx are Distributors — hyperscalers with existing cloud infrastructure, enterprise relationships, and the balance sheets to absorb a multi-year build-out. They are not betting on a single model. They are betting on owning the substrate.

Kantrowitz’s argument — as expressed in this episode — is that the pace of that commitment is itself the thing worth scrutinizing. When projected spend on buildings alone reaches $1.5T, the question the market must answer is: what revenue curve justifies that asset base?

That is not a hype question. It is a basic infrastructure economics question. And it is the right one to be asking.

AI CAPEX TRAJECTORY — PER KANTROWITZ’S ARGUMENT

Figures as cited by Alex Kantrowitz on Big Technology Podcast. Not independently verified by FourWeekMBA.

⚡ WHY THIS POINT CUTS THROUGH

Most AI debates center on model capability or market adoption. Kantrowitz’s framing shifts the lens to physical commitment — buildings, land, power contracts. That layer is slow to reverse. You do not pivot out of $1.5T in constructed infrastructure the way you depreciate a software license.

That asymmetry — between the speed of the capital commitment and the uncertainty of the return timeline — is the core analytical tension his argument surfaces. Whether it resolves as bubble or foundation depends on demand curves that are still forming.

The Bottom Line

A doubling of CapEx is a data point. A projected quadrupling — expressed in physical buildings — is a structural claim about what the world looks like in five years. Kantrowitz is not saying the bet is wrong. He is saying the scale of it demands a proportionate level of scrutiny. That is a reasonable ask.