The economy, at its core, is a closed loop.

Companies produce. Workers earn. Consumers spend. That spending becomes someone else's revenue. The cycle continues. GDP is the sum of all these exchanges, and roughly 70% of it in the US is internal consumer spending. It is about 59% in Canada and 55% in France to give a few examples. The economic engine in these countries runs on circulation of value between population.

Energy works the same way, actually. A power grid doesn't store much. It produces and consumes in near real-time. If production outpaces demand, the frequency in the grid rises and things break. If demand exceeds supply, you get blackouts. The system only works when the loop is balanced.

Now, what happens when you introduce a technology that massively increases output while simultaneously removing the consumers from the loop?

A research piece published this week by CitriniResearch models exactly this scenario. It's written as a fictional macro memo from June 2028, looking back at what they call the "Global Intelligence Crisis." The article is not a prediction. It's more of a stress test. And it's one of the most interesting articles I've read on the second-order effects of AI.

They speculate that: AI succeeds spectacularly and productivity soars. Margins expand and stocks rally. But the workers displaced by that productivity don't disappear from the economy, but they disappear from the spending side. And machines, however intelligent, don't buy groceries, pay mortgages, or take vacations.

The authors coined a term that stuck with me: "Ghost GDP" Output that registers in the national accounts but never circulates through the real economy. The factory is humming, but no one is in the store.

The feedback loop they describe has no natural brake. AI improves, companies need fewer people, displaced workers spend less, weaker demand pushes firms to cut costs further, which means more AI. In this scénario, each individual decision is rational, however the collective outcome is deflationary.

I've spent all my adult life building enterprise software. I've watched technologies come and go, some transformative, most overhyped. AI is different. I believe it's the most significant shift in how we build and operate businesses since the internet. But every previous wave of technological disruption had one thing in common: it created new roles that required humans. ATMs made branches cheaper, so banks opened more and hired more tellers. The internet killed travel agencies but created entire industries in their place and you certainly know of hundreds other examples.

The question that raises is whether AI breaks that pattern. Not because it can't create new jobs, it already has. But because it can also do those new jobs. The escape door we've always relied on with a new shiny technology may not be there this time.

When we automate both the earning and the spending side, the human is no longer in the loop, not to control, not to govern, not to correct course. This is no longer a technology problem. It's a humanity challenge.

And humanity challenges don't get solved by moving faster. They get solved by thinking clearly about what we're building, and for whom.

Inspired by The 2028 Global Intelligence Crisis — CitriniResearch, February 2026.