The Great Decoupling: How AI Is Rewriting the Labor Market

The global economy as we know it is dying.

Around the beginning of the 18th century, a “Great Divergence” began between ‘the West and the Rest’ – one that tipped the global socioeconomic scales for hundreds of years. 

It was an era when industrial innovation, among other things, saw European nations (and their colonized offshoots) enjoy strongly positive GDP growth as Eastern nations’ wealth shrank.

Thanks to inventions like the steam engine, power loom, and cotton gin, the West was able to quickly decouple industrial power from muscle. This shifted human workers’ “value” from their bodies to their brains. And since then, the global economy has functioned on a very simple, linear equation: Human Cognition + Time = Economic Value.

Whether we sought to build a bridge, write a legal brief, or design a next-gen semiconductor, we needed to pay for brainpower. That was the bottleneck.

We became a “knowledge economy.” We told our children that as long as they were smart, went to college, and learned to think, they would always have a seat at the table.

But as AI improves at breakneck speeds, we are now entering a new economic era: the Great Decoupling. 

And unlike the “Great Divergence,” where the ‘Rest’ struggled as the West thrived, this is a socioeconomic shift that will negatively impact us all…

AI Job Displacement and the New Era of Jobless Growth

In the first half of the 19th century, during the Industrial Revolution, we saw a phenomenon called Engels’ Pause. It was a 50-year window where the economy grew at record rates, corporate profits doubled, and the stock market soared – but wages for the average worker stayed flat or declined. Why? Because technology was now doing the heavy lifting; and the owners of that tech captured 100% of the gains.

In the Age of AI, Engels’ Pause 2.0 has begun. And this time, it’s on steroids. Just like everything else these days, it is bigger and moving faster than ever before. 

Today, the “Great Decoupling” is reflected in a cold, hard statistic: Corporate profits as a share of GDP have surged toward record highs of 11.55%, while labor’s share of the economic pie has plummeted to 53.8%, its lowest levels since the 1940s.

We are entering the era of “Jobless Growth.” Productivity is skyrocketing, but the gains aren’t going to the guy in the cubicle – they’re going to the guy who owns the algorithm. In the Gilded Age, we saw this with steel and rail. Today, we see it with “Centaur Startups.” For example, when Instagram was bought for $1 billion, it had 13 employees. Now AI-native startups are aiming for $100 billion valuations with a headcount you can count on two hands.

If you think a “Universal Basic Income” or a government handout is going to bridge this gap, you’re missing the point. The wealth is being concentrated in the Physical and Compute layers here. It’s being captured by the entities that own the data centers, the energy sources, and the silicon – not the laborers. 

The Myth of AI “Augmentation” – And What It Means for Workers

The mainstream media loves the word ‘augmentation.’ 

“AI won’t replace you; it will just make you more efficient.”

But if we’re being honest, “efficiency” is just a corporate euphemism for “we used to need five of you, and now we only need one.”

What we are seeing with models like OpenAI‘s Codex, Google‘s Gemini, or Anthropic‘s Claude Cowork isn’t just “better software.” It is Recursive Reasoning. 

These systems are now able to think through problems, check their own work, and iterate. When cognition becomes a commodity that can be downloaded for $20 a month, the “Knowledge Class” – lawyers, accountants, analysts, coders – becomes the new “Useless Class.”

If AI makes a lawyer five times more efficient, the world doesn’t suddenly need five times more legal briefs. It just needs 80% fewer lawyers.

This is the Arithmetic of Obsolescence.

It’s the reason the “Software-as-a-Service” (SaaS) dream is turning into a nightmare.

Look no further than Chegg (CHGG). It was the “knowledge toll booth” for students for a decade. Once ChatGPT arrived, that toll booth was bypassed entirely. The business model evaporated overnight, and the stock cratered more than 90%.

The giants aren’t safe, either. Salesforce (CRM) and UiPath (PATH) are built on “seat-based” pricing. But why pay for 1,000 “seats” of software when one AI agent can do the work of 1,000 employees?

The labor-replacement engine has been turned on, and it doesn’t have an “off” button.