Everyone’s talking about artificial intelligence. One of the calmest ways to understand it is simply to follow the money.

It can feel like the world has gone all-in on artificial intelligence overnight. The headlines swing between “this changes everything” and “this is a bubble.” Neither is very useful when you’re trying to think clearly about your own money.

So let’s step back from the noise and do something simple: look at how much is actually being spent, who is spending it, how it’s all being paid for, and where that spending is heading next. Understanding the AI investment cycle — and where we sit within today’s AI spending boom — really starts with that one question. The figures below come from research by Global X, a global investment manager, and from broader market estimates. They won’t tell you what to do — but they will help the picture make a lot more sense.

Where are we in the AI investment cycle? Earlier than you’d think

Big technology shifts tend to follow a pattern. A new idea arrives, money pours in to build it out, and over many years it reshapes the economy. We’ve seen it with the railways, with the internet, and with mobile phones.

Here’s the surprising part. Measured against the size of the world economy, the money going into AI today is still smaller than what went into those earlier build-outs at their peak — even though the eventual market for AI is estimated to be the largest of the lot.

The spending so far is real and large, but it hasn’t yet reached the levels earlier revolutions hit at their busiest. Global X reads that as a sign the build-out still has room to run — while noting that, as it matures, success will increasingly come down to which companies win the biggest share.

There’s another reason the companies keep spending: the potential prize is enormous. Global X estimates the future profits of the AI “platform” market could be worth around US$3.3 trillion in today’s money — far more than any single company has cumulatively spent building towards it so far. Independent research from McKinsey points the same way, describing the build-out as one of the largest investment cycles in modern history.

When the potential prize dwarfs the money spent chasing it, there’s a clear incentive to keep investing — which is exactly what we’re seeing.

Source: McKinsey — The cost of compute: a $7 trillion race to scale data centers

The spending is enormous — and still climbing

The clearest way to see the momentum is hyperscaler capex — the yearly capital spending, or AI capex, of the big cloud companies like Amazon, Microsoft, Google and Meta. These are the firms building the data centres, chips and systems that AI runs on.

The trend speaks for itself. Annual spending has climbed from around US$116 billion in 2020 to an estimated US$650 billion-plus by 2026. Broader market estimates (from FactSet and Goldman Sachs, as at June 2026) are higher still — consensus hyperscaler capex 2026 figures point to roughly US$757 billion in 2026 and US$920 billion in 2027, almost six times the level of just a few years ago.

Yes, the rate of growth is expected to ease from here — but the dollar figures keep rising. And it’s the absolute dollars that matter most for the long line of suppliers further down the chain: the makers of memory chips, networking gear and power equipment who fill those orders. Every extra hundred billion dollars of spend lands somewhere.

“The growth rate may be slowing — but the size of the cheque is still getting bigger every year.”

Source: International Energy Agency — Energy and AI (the scale of data-centre investment)

Who’s paying for it all? How the AI capex cycle is financed

For years, the big tech companies funded their spending out of their own enormous profits. That’s now changing — and it’s one of the most important parts of the story. As the cheques have grown, the cash these companies generate hasn’t kept pace: by the end of 2026, the largest cloud companies are on track to spend more on AI than they earn. To bridge the gap, they’re increasingly borrowing. This is the heart of AI capex financing the investment cycle — the question isn’t only how much is being spent, but where the money comes from.

Why does that matter to an everyday investor? Because borrowing changes the risk. Debt has to be serviced whether or not the investment pays off on schedule, which is why professional investors now watch company balance sheets and “credit spreads” closely. Breckinridge Capital Advisors notes these companies started from strong balance sheets, but that borrowing is rising and those spreads have begun to widen.

That leads to the real question hanging over the whole cycle: will the spending pay off? Right now, the money going out is running ahead of the new revenue coming in. Most analysts expect that to balance out over the next few years — but if returns keep lagging, market sentiment can turn quickly. It’s why AI-related shares can be bumpy even when the long-term build-out looks intact, and it’s the single thing most worth keeping an eye on.

Source: McKinsey — Who’s funding the AI data center boom?

Who has captured the value so far

Not every part of the AI story has been rewarded at the same time. Think of it as a relay race, where the baton passes from one group to the next.

First, the chip makers (companies like Nvidia and TSMC) surged as demand for AI chips took off — adding around US$1 trillion in value. Then the baton passed to the cloud giants, who gained over US$4 trillion as they turned AI into real services and revenue.

The part that, according to Global X, has not yet had its turn is AI infrastructure — the power, grid and data-centre companies that physically make all this possible. McKinsey maps this same chain — from chip and IT suppliers, to the cloud “operators,” to the power and equipment “energizers” further down the line. That’s the gap the next section is about.

Source: McKinsey — Who’s funding the AI data center boom? (the AI value chain)

Why “infrastructure” is the talk of 2026

All those data centres need one thing in vast quantities: electricity. And here lies the squeeze. Demand for power is racing ahead, while the grid that delivers it is growing only slowly.

Data-centre power demand is forecast to grow roughly 9–11% a year through 2030, while the grid that carries that power is set to grow only about 2.5% a year. On top of that, much of the grid in Europe and North America is decades old and due for replacement. The result is a growing need to build and upgrade the “plumbing” behind AI — which is exactly why power and grid companies are getting so much attention. Independent bodies see the same squeeze: the International Energy Agency expects data-centre electricity use to roughly double by 2030, and Goldman Sachs Research forecasts data-centre power demand rising as much as 165% over the same period.

Where the money actually goes

When you picture a data centre, you might imagine a big building. But the building itself is a surprisingly small part of the cost. Most of the money goes into what’s inside — the power and cooling systems that keep everything running.

Electrical and cooling systems together make up the lion’s share of the bill, with the computer equipment next, and construction and land trailing well behind. That’s a useful clue about where the spending flows — and it isn’t slowing down.

Globally, more than US$2 trillion is expected to be spent on data centres over the next five years alone — a scale echoed in Goldman Sachs research on the build-out. That’s the kind of sustained, structural data centre spending that tends to support a long build cycle rather than a quick fad.

Source: McKinsey — The cost of compute: a $7 trillion race to scale data centers

What this means for long-term investors

So what do we do with all of this? Calmly, and without overreacting in either direction, a few things stand out.

That last point is the one we care about most. At Plan My Wealth, our job isn’t to chase the latest headline — it’s to help you feel clear and confident about your money, whatever the markets are doing. If this has you thinking about the bigger picture, we’ve looked at why AI could shape your retirement over the decades ahead — and if you’d like a hand working out what it might mean for your own plan, that’s something we’re always happy to talk through — whether you’re in Melbourne’s north or elsewhere in Australia.

Source: McKinsey — Who’s funding the AI data center boom? (the “picks and shovels” of the AI value chain)

Frequently asked questions

Manny Tran GradDip (FinPlan), ABFP®, CRPC®

Director and Senior Financial Adviser

PLAN MY WEALTH

With 17+ years’ experience and over 1,000 retirement plans built for Australian families, Manny works with clients aged 50 to 65 across Bundoora, metropolitan Melbourne, and nationally via video consultation. His focus is helping pre-retirees replace uncertainty with a clear, evidence-based plan.

The Watermans Bundoora, Level 2, 1/3 Janefield Drive, Bundoora VIC 3083
+61 433 564 003 · manny@planmywealth.com.au · Book a free consultation