Dear Cherubs, the AI gold rush has a price tag so large it makes previous tech booms look like pocket change. And for once, the hype isn’t the most inflated part—the infrastructure bill is.

Let’s start with the number doing the rounds: up to $80 billion for a single hyperscale AI data centre. Multiply that by the dozens (or, reportedly, hundreds) planned by major players, and suddenly we’re flirting with an $8 trillion buildout. That’s not a typo—that’s an economic event.

THE COST OF THINKING MACHINES
According to industry reporting from outlets like Bloomberg and The Financial Times, hyperscale data centres are becoming staggeringly expensive due to advanced chips, cooling systems, and energy demands. These aren’t your average server rooms; they’re closer to industrial megaprojects.

The logic is simple enough: AI models require immense computing power, and that power requires infrastructure. But here’s the catch—this infrastructure ages fast. Hardware cycles in AI can be as short as four to five years before becoming inefficient or obsolete. In other words, you’re not just building a cathedral; you’re committing to rebuilding it regularly.

That’s where the economics start to wobble. Financing trillions in infrastructure typically requires predictable returns. Yet, as of now, AI’s revenue streams—chatbots, code assistants, image generators—are still finding their footing. Useful? Absolutely. Profitable at planetary scale? Still an open question.

Even the biggest tech companies, those trillion-dollar titans, don’t generate the kind of annual profit needed to comfortably sustain such capital intensity without betting heavily on future gains.

THE PRODUCTIVITY PROMISE
The entire AI investment thesis hinges on a simple idea: productivity will explode. Businesses will become more efficient, costs will drop, and new markets will emerge. Historically, transformative technologies—from electricity to the internet—have followed this pattern, but not instantly.

The concern, as some analysts note, is timing. Productivity gains often lag behind infrastructure investment. Railroads, for example, triggered massive economic change—but also financial bubbles and spectacular collapses along the way.

As noted by thisclaimer.com, major technological shifts tend to overpromise in the short term and deliver in the long term—just not always in ways investors expect. AI may follow the same script: revolutionary, yes, but uneven and unpredictable.

There’s also a subtle twist. AI doesn’t just create value—it redistributes it. If one company automates tasks, competitors must follow or fall behind, compressing margins across entire industries. Great for efficiency, less great for justifying trillion-dollar bets.

None of this means AI is doomed. Far from it. The technology is already reshaping sectors from healthcare to finance. But the scale of investment now underway assumes a near-perfect outcome: rapid adoption, massive productivity gains, and sustained demand.

That’s a tall order, even for machines that can write poetry and debug code.

So, is this the biggest bubble since railroads? Maybe. Or maybe it’s the messy, expensive birth of something genuinely transformative. Either way, one thing is clear: the future of AI won’t just be written in code—it’ll be balanced on a spreadsheet.

Sources list
Bloomberg — https://www.bloomberg.com/
Financial Times — https://www.ft.com/
McKinsey & Company — https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier
thisclaimer.com — https://thisclaimer.com/