The Ouroboros Economy: How Corporate America, Wall Street, and AI Companies Are Eating Their Own Customers
They’re optimizing themselves out of a market.
In 1914, Henry Ford did something that scandalized the business world. He doubled his workers’ wages to five dollars a day and cut the workday to eight hours. The establishment called it madness. Ford called it survival. He understood a principle so elementary it shouldn’t need restating: the people who build your products need to be able to buy them. Work design is consumption design. The factory floor and the showroom floor are connected by a paycheck.
More than a century later, Corporate America, Wall Street, and the artificial intelligence industry are conducting a live experiment in what happens when you forget that lesson — or worse, when you decide it no longer applies.
The Math That Nobody Wants to Do
The American economy is a consumer economy. That is not an opinion. It is an accounting identity. Consumer spending represents approximately 68–70% of U.S. GDP, according to Bureau of Economic Analysis data through late 2025 — above the long-term historical average of 64.4%. Personal consumption expenditures reached roughly $19.7 trillion in the fourth quarter of 2025 alone, the highest quarterly figure ever recorded. Every dollar of that spending traces back to someone with income, and most of that income traces back to a job.
Now consider what is happening to those jobs.
The World Economic Forum’s 2025 Future of Jobs Report, which surveyed more than 1,000 leading employers across 55 economies, projects that 92 million jobs will be displaced by 2030. While the report anticipates 170 million new roles being created — yielding a net gain of roughly 78 million — the transition period is where the damage lives. Forty percent of the employers surveyed said they plan to reduce their workforce as AI automates tasks once done by humans. Half said they plan to fundamentally reorient their business models around artificial intelligence.
Meta’s March 2026 announcement made this concrete. The company cut approximately 16,000 jobs — roughly 20% of its workforce — and explicitly linked the reductions to a $600 billion AI capital expenditure plan running through 2028. This was not a company in financial distress trimming fat. This was a profitable enterprise replacing human labor with machines, tracked and measured by internal AI utilization dashboards. As one analysis observed, productivity gains from these moves flow to corporate profits and shareholder wealth within quarters, while workers absorb losses on a delayed schedule of two to four quarters. The asymmetry is the feature, not the bug.
Amazon eliminated 14,000 corporate roles, stating that AI enables leaner structures. Workday cut 8.5% of its workforce — about 1,750 people — to redirect resources toward AI investments. Nearly 55,000 job cuts in 2025 were directly attributed to artificial intelligence, according to the outplacement firm Challenger, Gray & Christmas, out of a total 1.17 million layoffs — the highest level since the pandemic year of 2020.
Here is the question nobody in the C-suite seems to be asking: Who do they think is going to buy their products?
The Demand Destruction Feedback Loop
A company that eliminates fifty thousand jobs saves on labor costs in the short term. Analysts applaud. The stock spikes. But if that pattern scales across industries — and every earnings call in America is promising exactly that — the aggregate effect is a contraction of the consumer base. Fewer people working means fewer people spending. Fewer people spending means demand destruction. Demand destruction means revenue pressure. And revenue pressure means the margin gains from those layoffs erode.
This is not speculative economics. This is the logic that every first-year MBA student is taught, then apparently forgets the moment they sit in a boardroom.
The K-shaped economy is already visible in the data. Research from the Federal Reserve Bank of Boston, using large-scale credit card data, found that since 2022, high-income consumers have been the primary engine driving aggregate spending growth. Real spending by the lowest-income consumers increased sharply during the pandemic stimulus period and has grown only modestly since, while spending by top earners has remained robust. The top 20% of earners now account for approximately 57% of total U.S. consumption, up from 53% in the 1990s, according to analysis from the Federal Reserve Bank of Dallas.
Meanwhile, household debt has reached $18.8 trillion as of Q4 2025. Credit card balances swelled to $1.23 trillion. Student loan serious delinquency rocketed from 0.70% to 16.19% in a single year — not a gradual deterioration but a cliff — as pandemic-era protections ended and roughly one in four federal student loan borrowers fell behind on payments. Low-income consumers now carry substantially higher credit card debt than they did before the pandemic. The bottom half of the economy is not “spending” in any meaningful sense. It is borrowing to survive.
The practical implications are already showing up in corporate earnings. McDonald’s reported losing traffic from lower-income consumers who can no longer comfortably afford fast-food meals. When a significant share of your customer base can’t afford a Big Mac, you have a demand problem that no amount of AI-powered drive-through optimization will solve.
The B2B Mirage
A common counterargument holds that even if consumer-facing businesses contract, business-to-business companies — enterprise software, cloud infrastructure, AI tooling — can sustain growth through corporate demand alone. This is a comforting fantasy with an expiration date.
B2B revenues ultimately depend on the vitality of end markets. Manufacturers purchase from suppliers based on anticipated consumer orders. Service firms rely on retail clients whose sales forecasts derive from consumer activity. Cloud providers scale in response to their clients’ expectations of increased business volume. As the 2026 Report on Employer Firms from the Federal Reserve’s Small Business Credit Survey noted, nearly half of businesses rely on international suppliers, and most have experienced rising input costs year over year. Revenue and employment growth remained steady between 2024 and 2025, but firms were slightly more likely to report a decrease in revenues — an early signal that the downstream consumer weakness is beginning to migrate upstream.
The entire supply chain is, in the end, a chain. Pull the consumer out from the bottom and the whole thing unravels — slowly at first, then all at once.
The $660 Billion Question
In the meantime, the AI investment boom thunders forward. Businesses collectively spent an estimated $410 billion on AI in 2025, and projections for 2026 call for $660 billion more. Goldman Sachs analysts have escalated their rhetoric, claiming that AI has had effectively zero measurable impact on U.S. economic growth over the course of 2025. Economists at the National Bureau of Economic Research have identified what they call a “productivity paradox,” in which companies perceive large productivity gains from AI, but those gains fail to materialize in measurable economic output — likely reflecting delays in translating efficiency into actual revenue.
Dario Perkins, head of macroeconomics at TS Lombard, put it bluntly in the Financial Times: while U.S. productivity has been strong and hiring weak, the evidence suggests cyclical forces rather than genuine automation-driven transformation are responsible.
So we have a situation in which hundreds of billions of dollars are being funneled into a technology whose primary demonstrated use case, so far, has been eliminating the paychecks of the people who make up 70% of GDP. The investment generates returns for capital owners immediately. The consumer damage arrives on a lag. And by the time the lag catches up, the structural erosion of purchasing power may be difficult to reverse.
Wall Street’s Role: The Buyback Machine
The AI story does not exist in a vacuum. It exists within a broader pattern of corporate capital allocation that has systematically prioritized shareholders over workers for decades — a pattern that AI is now supercharging.
Stock buybacks were projected to hit a record $1 trillion in 2025, according to Americans for Financial Reform. Over the preceding fifteen years, approximately 94% of corporate profits went to shareholders in the form of buybacks and dividends rather than to workers, according to research from Inequality.org. In the two months following the 2017 Tax Cuts and Jobs Act, S&P 500 companies directed 106 times more money to share buybacks ($160 billion) than to wage increases ($1.5 billion).
The scale is staggering. The Institute for Policy Studies found that the 100 S&P 500 companies with the lowest median worker pay spent $522 billion on stock buybacks over five years. Nearly half of these companies spent more on buybacks than on capital investment. Google authorized an additional $70 billion in repurchases. Meta expanded its buyback program by $50 billion. NVIDIA announced $60 billion in buybacks. JPMorgan authorized $50 billion in July 2025.
Every dollar spent on buybacks is, by definition, a dollar not spent on wages, training, research, or long-term productive capacity. Stock-based compensation makes up roughly 80% of CEO pay, and buybacks artificially inflate share price — creating a closed loop in which executives personally benefit from the very decisions that hollow out their workforce. The SEC has confirmed that executives regularly time personal stock sales to coincide with the price surges that follow buyback announcements.
This is not capitalism. This is extraction.
The Tax Code Is Rigged for Machines
The structural incentives make the problem worse. Under current bonus depreciation rules, a business can immediately expense the full cost of an AI server. But investing in the humans who need to learn how to work alongside that technology? Six major provisions in the Internal Revenue Code create bottlenecks that effectively penalize workforce retraining. A business cannot even deduct the cost of training an employee to meet the minimum requirements of their current job.
The tax code, in other words, actively subsidizes replacing people with machines while making it more expensive to keep people employable. This is not a market failure. It is a policy choice — one that no serious legislative effort has yet attempted to reverse.
What “Losing the Plot” Actually Looks Like
To “lose the plot” is to become so absorbed in the details of execution that you forget what the story is actually about. In the case of corporate America and the AI industry, the story is supposed to be about creating value — for customers, for workers, for society. Instead, the plot has become about creating efficiency for its own sake, about impressing Wall Street with headcount reductions, about hitting quarterly EPS targets by eliminating the very humans who constitute your market.
The shareholder value argument assumes a static market. Markets are not static. They are made of people. And people need income to participate in them.
What we are witnessing is not bold leadership. It is the absence of it. It is what happens when executives optimize for one variable — short-term labor cost — without modeling the system that variable lives inside. Nobel laureate Daron Acemoglu has spent years documenting how automation that displaces workers without creating comparable new opportunities leads not to broad prosperity but to concentrated wealth and declining social mobility. The data, from Keynes to Brynjolfsson to Autor, is unambiguous: wages and purchasing power are the foundation of aggregate demand. Undermine them, and you undermine the economy itself.
Henry Ford grasped this in 1914. He wasn’t a philanthropist. He was a pragmatist. He understood that the assembly line only works if the people running it can afford the car at the end of it. The labor-consumption loop was not an accident of the industrial age. It was the architecture that made the industrial age possible.
We are now dismantling that architecture at machine speed, without replacing it with anything that can sustain the economy it built.
The Way Forward, If Anyone Wants It
There is no version of a thriving economy in which AI generates massive productivity gains and those gains flow exclusively to a small ownership class while 70% of GDP slowly starves for oxygen. The math doesn’t work. It has never worked. Every time capital has tried to decouple from labor — the Gilded Age, the run-up to the Great Depression, the financial crisis of 2008 — the correction has been painful and, in some cases, catastrophic.
The solutions are not mysterious. They include rebalancing the tax code so that investment in human capital is at least as attractive as investment in machines. They include rethinking corporate governance so that quarterly EPS is not the sole measure of success. They include policies that ensure the productivity gains from AI are shared broadly — through higher wages, shorter workweeks, portable benefits, and genuine retraining infrastructure — rather than captured entirely by shareholders and executives.
They include, above all, a recognition that the economy is not an abstraction. It is a network of human beings producing and consuming. Sever the link between production and consumption, and you do not get a more efficient economy. You get a smaller one.
The AI companies building these tools, the corporations deploying them, and the Wall Street analysts cheering them on are not, in most cases, acting out of malice. They are acting out of a failure of imagination — an inability or unwillingness to see that the system they are optimizing is the same system they depend on. They are the ouroboros, the snake eating its own tail, convinced that each bite is nourishment.
Henry Ford’s lesson is more than a century old. It has never been more urgent. The question is whether anyone with the power to act on it will do so before the feedback loop closes — and the customer base that built Corporate America is too hollowed out to sustain it.
Brian Curry is a Kansas City–based data scientist, AI researcher, and founder of Vector1 Research, an independent lab advancing marketing economics, causal inference, and cognitive AI systems. He has led data and analytics initiatives at enterprises from $4B to $125B, including Koch Industries, Tractor Supply, Vail Resorts, Hallmark, Garmin, and AT&T. He is the creator of MeaningFlow (semantic content modeling), Papilon (causal inference and complex systems), PyCausalSim (causal discovery through simulation), and the Memory-Node Encapsulation (MNE) architecture for cognitive AI.
Contact: brian at vector1.ai | Vector1 Research | GitHub
Sources and references: World Economic Forum Future of Jobs Report 2025; Bureau of Economic Analysis PCE data (Q4 2025); Federal Reserve Bank of Boston credit card research (Hagler and Patki, 2025); Federal Reserve Bank of New York Household Debt and Credit Report (Q4 2025); Goldman Sachs AI productivity analysis (2026); National Bureau of Economic Research productivity paradox working paper; Americans for Financial Reform buyback analysis; Institute for Policy Studies Executive Excess Report 2024; Challenger, Gray & Christmas layoff data (2025); Federal Reserve Bank of Dallas consumption research; UC Berkeley service sector multiplier methodology.
