We may be approaching an inflection point that challenges the foundational logic of the U.S. economy. A sort of elastic limit that the economy can bear without needing to evolve into something completely different.

And guess what? Some have seen this movie before.

The Setup

Consumer spending drives roughly two-thirds of U.S. GDP. It’s a simple equation: employed workers earn wages, spend those wages, and that spending fuels corporate revenue and subsequent hiring. A virtuous cycle…until it isn’t.

The Manufacturing Precedent

Between 1950 and 2019, manufacturing employment in the US collapsed from 25% of non-agricultural jobs to just 8%. From 2000 to 2024 alone, 4.5 million manufacturing jobs vanished. Yet industrial production grew 77% between 1980 and 2019 as employment fell 27%.

The pattern was clear: automation eliminated low-skilled repetitive work while creating demand for elite operators, programmers, and engineers. The displaced factory worker earning $50,000 wasn’t retrained to become the $120,000 robotics engineer. Different people filled those roles; people with aptitude, education access, and often existing economic advantages to acquire these new skills.

Manufacturing’s decline was cushioned by a safety valve: the professional services economy absorbed displaced workers into middle-tier white-collar work. HR coordinators, logistics analysts, customer service managers; roles requiring judgment and coordination that seemed automation-resistant…at the time.

Fast-forward to the Great AI Disruption: The Safety Valve Closes

The $400 billion corporate investment in AI isn’t just another technology upgrade. Amazon’s recent elimination of 30,000 white-collar positions across e-commerce, HR, and logistics signals something more fundamental: we’re now automating the automation refuge.

These aren’t manufacturing jobs. These are precisely the roles that absorbed the previous generation’s displaced workers; the analytical, coordination, and judgment-based work that seemed immune to technological displacement.

The Economic Contradiction

Here’s where the math gets uncomfortable:

• AI drives corporate efficiency; Profits multiply; Shareholder wealth concentrates
• AI displaces workers; Earning power declines; Consumer spending contracts
• Consumer spending shrinks: Corporate revenues pressure; Economic growth slows

We’re engineering a system where productivity gains and profit growth become decoupled from broad-based prosperity.

The Wealth Concentration Accelerant

From the 1950s through early 1980s, manufacturing automation concentrated wealth in two groups: capital owners and the elite technical workforce. But those high earners still spent money, creating demand for retail associates, food service workers, personal services. The wealth circulated, even if unequally.

By the mid-1980s, even this circulation began to weaken as offshoring and advanced automation accelerated job losses faster than elite technical roles could absorb displaced workers.

AI is replicating the concentration pattern at scale, but with two critical differences: first, there’s no obvious “next rung” for displaced workers to climb down to. Second, AI doesn’t have disposable income. The productivity gains flow directly to shareholders as profit, not back into the broader economy through consumption. The circulation breaks.

The beneficiaries of this transformation are narrow and predictable:

1. Existing shareholders in AI-leading companies, wealth compounds on existing wealth
2. Elite technical workers, AI researchers, advanced data scientists, systems architects whose skills complement AI rather than compete with it
3. Capital owners who can deploy resources into AI infrastructure and applications

The manufacturing analogy is instructive but ominous. The displaced factory worker didn’t retrain into robotics engineering. Children of some factory workers, if they had access to education, stable home environments, and early technical exposure, became engineers.

But most didn’t. The skill stratification was permanent and generational.

The Elite Skills Trap

AI follows the same pattern, but accelerated and amplified:

• Mid-level data analyst roles could evolve to elite ML engineers and AI researchers
• HR coordinators roles could evolve to people analytics specialists with advanced statistical modeling skills
• Financial analyst roles already evolving to quantitative developers that help train AI on financial algorithms

Each transition demands not just “retraining” but fundamental cognitive aptitude, years of specialized education, and, critically, the economic stability to pursue that education.

The $145,000 logistics manager laid off at 45 with a mortgage and kids in school isn’t well-positioned to start a new career after spending four years getting a PhD in computer science.

Meanwhile, those with existing advantages, including educational pedigree, financial cushion, early exposure to technical fields, are indeed well-positioned to capture the high-value roles that AI creates. Wealth concentrates not just with capital returns but in a cognitive elite whose skills remain valuable precisely because they’re rare and difficult to acquire.

The Consumer Economy Question

If consumer confidence is already at six-month lows with early recession signals emerging, what happens when this displacement accelerates?

Manufacturing’s decline from 25% to 8% of employment was gradual enough, with the benefit of a white-collar economy as a landing pad, that the consumer economy could adjust. Professional services grew, the middle class shifted sectors but remained intact enough to drive consumption.

This time, we’re automating the landing pad itself. And the timeline is compressed.

The wealthy spend, but they don’t spend like the middle class. A household earning $100,000 spends most of it on housing, goods, services; discretionary purchases that flow through the economy. A household worth $10 million doesn’t proportionally increase their consumption. They invest, they save, and they concentrate wealth.

Therein lies the paradox: we’re optimizing for corporate profitability in an economy structurally dependent on mass earning and spending power.

The Uncomfortable Truth

Manufacturing automation taught us that productivity gains and job losses can coexist for decades. Production rose 77% while employment fell 27%. Shareholders prospered. A technical elite emerged. And millions of workers permanently lost economic ground.

But manufacturing was only 25% of employment at its peak. Professional services, the sector now facing AI displacement, represents a much larger share of the workforce and an even larger share of consumer spending power.

The policy conversations ahead likely need to move beyond “jobs retraining” platitudes. Manufacturing proved that displaced workers don’t retrain into elite technical roles; rather, different people, with different advantages, fill those positions.

We need to grapple with harder questions:

• How do we distribute productivity gains when labor’s bargaining power is structurally diminished, and the skill premium becomes increasingly steep?
• What does a consumer economy look like when both capital returns and high-skill wages concentrate among those already advantaged?
• Can we sustain consumer spending when the professional middle class, the economy’s spending engine, faces systematic skill devaluation?
• Are we willing to reimagine social contracts around work, income, and value creation, or will we repeat manufacturing’s pattern at a scale that breaks the consumer economy model entirely?

The Bottom Line

The AI transformation isn’t a future scenario, it’s the present reality, reshaping employment, earnings, and economic structure. The companies making $400 billion in AI investments are rational actors optimizing for their shareholders.

Manufacturing automation showed us the playbook: productivity rises, employment falls, wealth concentrates in capital and elite skills, and the displaced don’t retrain into the elite roles, they ratchet down economically, often permanently.

The difference this time? Manufacturing was a sector. Professional services is the economy. And we’re running out of “next rungs” for people to climb down to.

The question isn’t whether AI will boost productivity, it will. The question is whether our economic system can remain viable when those productivity gains systematically undermine the consumer spending base the entire system depends upon, and when the skill requirements for remaining valuable become increasingly concentrated among those with aptitude, education access, and existing economic advantages.

We’re running an experiment in real-time, with very little contingency planning around potential scenarios and long-term implications.

Question remains: What mechanisms, if any, prevent this from breaking the consumer economy model entirely?