The AI infrastructure boom is here, but will it last

Five charts to start your day

Let’s start with this: big bet, big buildout, yet big question marks. When Nvidia released its latest earnings and guidance, with quarterly revenue jumping sharply year-on-year and an even stronger outlook for the months ahead, markets responded with enthusiasm. The company’s momentum has become a kind of barometer for the entire AI cycle. But its success tells only part of the story.

Across these five charts one tension stands out. The AI buildout is accelerating fast, shown by soaring imports of data centre equipment. Yet the equity rally that sits on top of it is still modest by bubble standards, despite the noise. To keep pace, big tech is borrowing heavily, issuing debt at a rate that dwarfs recent years. Credit investors are reacting first, pushing spreads wider for companies that used to trade as near risk free. At the same time the corporate conversation is shifting from chips and data centres to AI agents, a sign that the market wants applications and revenue, not just infrastructure. The machines are going in at speed, but the returns they are meant to produce are still uncertain.

Nvidia’s strong results help explain why the buildout continues at this pace. They give the sector confidence and create the sense that the investment is justified. Yet they also sharpen the question sitting underneath the entire story. If one company is capturing so much of the upside, what risks remain for the rest of the ecosystem? And what happens when the focus moves from supply to outcomes? The scale is impressive, but the payoff is still unproven.

These charts show what words cannot.

Quick note — Killer Charts has been $8/month and $80/year since launch two years ago. From 1 December the price for new subscribers will rise to $10/month and $100/year. If you’ve been meaning to upgrade, you can lock in the old rate for life here.

CHART 1 • The AI infrastructure boom is here, but will it last

What this chart shows is how imports of data centre equipment have jumped dramatically, while other equipment imports have risen more modestly.

This explosive growth represents just another way to understand the infrastructure demands of artificial intelligence. While the dotcom bubble saw steady increases in tech infrastructure, today’s AI revolution is driving imports at nearly twice the historical rate. The gap between data centre equipment and other imports has never been wider, reflecting the extraordinary computing power required to train and run large language models. Unlike previous tech cycles, this isn’t just about connectivity or storage. It’s about raw processing capability on a scale we’ve never seen before.

Is this infrastructure buildout sustainable, or are we witnessing the early stages of overcapacity?

Source: fDi Intelligence

It is amazing how quickly confidence can splinter. Money continues to pour into AI infrastructure yet the conversation is already shifting to usefulness and execution.

It is also a reminder that progress never moves in a straight line. Markets wobble long before headlines change and investors often sense strain before it is obvious in the income statements. The AI story is still unfolding and the gap between expectation and evidence is where the real risk sits.

I have got four more charts that expand on this story but they are for paid subscribers. Consider joining if you want the full edition.

---