Remember this? One day Henny-penny was picking up corn in the rickyard when—whack!—an acorn hit her upon the head. “Goodness gracious me!” said Henny-penny, “the sky’s a-going to fall; I must go and tell the King.”
Every few months, Henny-penny surfaces around the AI argument:
AI is too expensive.
The infrastructure can’t keep up.
The math doesn’t work.
The bubble is about to burst.
The latest take[1] frames it as something more dramatic — that the economics of AI are starting to “come crashing down.”
That’s not what’s happening. What we’re seeing is something far more familiar — and far more important.
The Misread: Treating AI Like a Product Instead of Infrastructure
Most of these arguments share the same flaw. They evaluate AI as if it’s a standalone business that needs to justify its cost directly.
Cost per token.
Revenue per model.
Margins at the platform level.
That’s a clean way to analyze a SaaS tool. It’s the wrong way to analyze something that behaves more like infrastructure – something that cross across business sectors. Consider these behavior patterns that emulate AI:
Electricity didn’t make sense at first.
Cloud didn’t make sense at first.
Even the internet didn’t make sense at first.
All of them looked “too expensive” when measured in isolation. All of them became indispensable once they started reorganizing how work actually gets done…how their behavior patterns cut across sectors and had a profound effect on the way people work and what they do.
AI is following such a path.
Yes, The Costs Are Real
Of course the beginnings cost money. What “start” doesn’t? However, there’s no need to dismiss the pressure points.
Compute is expensive.
Data centers are constrained.
Power is becoming a gating factor.
And usage is exploding — especially with agents.
The “free ride” phase might be ending, but as any business person will tell you, there is no such thing as “free.” Everything has a cost. And, as pricing starts to show its ugly head, access pressures start to emerge. Questions are asked. But none of this signals collapse.
What it does signal is transition from subsidized growth to economic discipline. Just the way electricity, the cloud and the internet did.
That’s not a collapse. That’s maturity.
The Bigger Miss: Where Value Actually Shows Up
If you Google “Should I Lower My Price,” Google delivers (including in his AI evaluation) my blog from a few years ago: Never Lower Your Price. Always Increase Your Value. In that piece I argued: “The price of anything is always tied to perceived value. Going low (or high) is a matter of perceived value – always. The inherent value of anything is not in the thing: it’s in the perception of the buyer.”
Value is one of those great intangible words that everyone throws around, but no one defines.
Bob Stone once said that people buy anything for one of two reasons: to protect what they have or to gain something. In both answers, value is imbedded in the decision. If value is perceived, the purchase is made. If not, it’s not.
That’s what you see in AI today: both reasons at work.
The collapse narrative saying the sky is falling assumes that AI has to justify itself directly through its own pricing. That’s not where its value is being created.
AI’s value is showing up in unanticipated places the model doesn’t measure well or account for:
Fewer steps in a workflow.
Fewer people required to complete a task.
Fewer errors.
Faster decisions.
Tighter coordination across systems.
In other words: AI isn’t just consuming resources. It’s replacing resources – in many cases, more expensive ones. Things like time, labor, reworks, delays.
If you’re only looking at token costs, you’re missing the equation entirely.
The $2 Trillion Problem (That Isn’t )
When you see projections that the industry needs trillions in revenue to justify current investment, you see the sky falling. It’s not. Those numbers sound alarming, but they’re also built on a fragile, mistaken assumption: that today’s AI cost structure is static.
History doesn’t support that. Computing gets cheaper, infrastructure gets more efficient and software gets better at using less. Every major technology wave followed that curve. There’s no reason to believe AI won’t.
What’s Actually Happening
No, the sky isn’t falling. It just looks a little cloudy, like all storms look like before it rains. What we are witnessing is actually a compression point of three forces converging all at once:
- Infrastructure constraints – the demand for data centers is enormous as I am sure you have heard. The megaprojects in construction reports for these facilities is at all time highs. According to ConstructConnect™ who knows these things, the biggest YTD dollar gains have come from increased Offices which include data centers starts, which are now at $48.9 billion, representing a gain of $38.7 billion and a near quadrupling of comparable starts from the same period in 2025.
- Exploding demand. Some relevant statistics: 77% of companies are currently using or exploring AI, with 83% listing it as a top business priority. AI computational performance has increased a thousandfold since 2018. And, individual employees using AI tools report productivity boosts of up to 40% to 80%
- Pricing normalization. The federal reserve of San Francisco has a 100+ page paper on The Rise of AI Pricing: Trends, Driving Forces, and Implications for Firm Performance. Among other things, they noted “Firms that adopted AI pricing experienced faster growth in sales, employment, assets, and markups, and their stock returns are also more responsive to high-frequency monetary policy surprises than non-adopters.”
All this creates tension. But tension doesn’t break systems like AI. Tension forces AI to reorganize into something else — better pricing models…more efficient usage and clearer segmentation between enterprise and consumer value.
In other words — a market starting to behave like a market.
The Real Shift (And Why It Matters)
If you want to understand where AI is going, stop asking: “Does AI pay for itself?” Instead, start asking “What happens to a business that learns how to build around and with it?”
Because that’s where the divide is forming. Not between companies that use AI and those that don’t. But between those who treat it as a tool and those who use it to reshape their workflows. That’s where margin moves. That’s where advantage builds.
Bottom Line?
No, the sky isn’t falling. The economics of AI aren’t collapsing. It’s just getting ready to rain opportunities. And like every storm before it, storms won’t break everything – just the weaknesses. Storms like AI reshape our movement in business.
So, the question isn’t whether the AI model works. The real question is who will learn how to work within AI. Hopefully, that’s all of us! Let’s hear your opinion!
____________________________________________________________
[1] The Horrible Economics of AI Are Starting to Come Crashing Down, By Victor Tangermann Published Apr 24, 2026