The Gap Between The US Stock Market & The Economy
Wall Street’s Record Run and the Economy Underneath It
Just yesterday the S&P 500 crossed 7,500 for the first time in its history. The Dow Jones Industrial Average simultaneously breached 50,000, a level it had not seen since February, when the Iran war began. Technology stocks led both moves, with Nvidia adding 4.4 percent in a single session after Washington cleared ten Chinese firms to receive H200 chips during President Trump’s Beijing summit. Cisco Systems jumped 13.4 percent after lifting its revenue and earnings guidance. The Nasdaq hit an all-time high at 26,635. By any headline measure, American capitalism was having a spectacular week.
Buried in the same week’s data was a different set of numbers. The Bureau of Labor Statistics had released the April employment situation on May 8. The report showed nonfarm payrolls rising 115,000, unemployment holding at 4.3 percent. The headline beat a weak consensus forecast of 62,000, which should have been encouraging. The household survey told a different story. Within the same release, 358,000 Americans moved into fresh unemployment of less than five weeks, and 445,000 workers shifted to part-time status for economic reasons: 803,000 people moving into active labor distress in a single month, a figure that received no headline treatment. Chicago Fed President Austan Goolsbee, interviewed on CNBC the morning of the release, said the labor market had been “pretty much stable for a year, year and a half.” The Bureau of Labor Statistics made the same point in its own text: payrolls had shown “little net change over the prior 12 months.” February’s revised payroll figure was negative 156,000.
The S&P 500 added $55 to its level that week. The labor market printed twelve months of effectively no net growth. Both of these things are true simultaneously, and neither cancels the other out. Understanding why requires rejecting the premise that the stock market is a report card on the American economy. It is not. It is a report card on American corporate ownership, and those two things separated decisively some time ago.
What the Index Actually Measures
The S&P 500 is presented in every financial broadcast as the barometer of economic health. The framing is so embedded that politicians cite it as evidence of policy success and journalists use it to set context for consumer sentiment. What the index actually measures is the aggregate market capitalization of five hundred large companies, weighted by their size, which means its performance reflects the fortunes of a narrow ownership class rather than the economy those companies operate inside.
This distinction is not semantic. As of the fourth quarter of 2025, the wealthiest one percent of American households held fifty percent of all U.S. stock market wealth, approximately $29 trillion, according to Federal Reserve data compiled by The Motley Fool. The top ten percent held eighty-seven percent of stocks, $44 trillion. The bottom fifty percent of American households held one percent of stock market wealth: roughly $620 billion, divided among approximately 65 million households, averaging under $10,000 each. That average is misleading because the distribution within that bottom half is deeply unequal. The median holding is close to nothing.
The index’s concentration problem has compounded over time, and accelerated sharply since the AI trade took hold. The eight largest companies in the S&P 500 accounted for 35.6 percent of the entire index at the end of 2024, compared to 15.5 percent a decade earlier, per S&P Dow Jones Indices data. In December 2024, the index’s 0.06 percent total return gain would have been a loss of 0.18 percent without the Magnificent Seven. In the first quarter of 2026, 84 percent of S&P 500 companies reporting beat earnings per share estimates at a rate nearly three times the historical average, per FactSet’s weekly earnings update of May 8. The forward twelve-month price-to-earnings ratio has risen to 21.0 against a ten-year average of 18.9 and a five-year average of 19.9. The market is priced to a version of the future that requires AI’s promised productivity gains to materialize, at scale, within the next several quarters.
The index, in short, has become a concentrated bet on a handful of companies whose collective business case rests on a single transformative thesis that the data, at this stage, has not yet confirmed.
The Buyback Mechanism and the EPS Illusion
To understand how corporate America has managed to report record profit margins while simultaneously shedding workers, it is necessary to understand share repurchases not as a financial mechanism but as a political one.
The basic arithmetic: when a company buys back its own shares on the open market, it reduces the total number of shares outstanding. Earnings per share, by definition, is total earnings divided by total shares. Reduce the denominator without changing the numerator and EPS rises, not because the business earned more money but because the math got manipulated. The stock price, which analysts and investors track against EPS, rises accordingly. Executive compensation, tied overwhelmingly to stock performance through options and restricted stock units, rises with it. The incentive to reduce headcount, which cuts costs and thus nominally increases total earnings, compounds the mechanism.
In 2024, S&P 500 companies spent $942.5 billion buying back their own shares, a new annual record, according to S&P Dow Jones Indices. Q1 2025 set an additional record for a single quarter at $293.5 billion. Combined dividends and buybacks in 2024 totaled $1.572 trillion, per S&P DJI’s official tally, in total shareholder returns. Apple alone distributed more than $100 billion in dividends and buybacks in 2024, per Axios reporting. The Information Technology sector accounted for 26.2 percent of all S&P 500 buybacks in Q4 2024. These are not incidental corporate housekeeping decisions. They represent the primary mechanism through which the stock market’s price appreciation is manufactured independently of underlying economic conditions.
The Carson Group’s analysis of S&P 500 total return composition from 2020 through mid-2025 quantified this precisely. Forward profit margins expanded from under twelve percent before the pandemic to a record 14.0 percent by mid-2025. Decomposing that expansion reveals that it came from two sources: nominal GDP growth, which was partly inflationary rather than reflective of genuine output gains, and operating leverage, meaning companies extracted more earnings from their existing revenue base by shrinking their cost structures. The primary cost structure being shrunk was labor. Between 2000 and 2019, per S&P Dow Jones analysis, interest rate declines and successive rounds of corporate tax reduction drove more than eighty percent of S&P 500 margin expansion. The current margin record has been built on the same two inputs, augmented by workforce reduction.
At the same time, median total compensation for U.S. S&P 500 CEOs reached $17.1 million in 2024, a 9.7 percent increase year over year, per Equilar data cited by the AFL-CIO. The ratio of CEO to median worker pay reached approximately 285 to one in 2024. To put that ratio in economic terms: for every dollar a median employee earned, the CEO of the same company earned $285. The margin expansion that drove the stock price rise was funded, partly, by the gap between those two numbers widening.
Microsoft provides the cleanest case study. In May 2025, the company reported quarterly earnings that beat analyst expectations on both revenue and profit. Weeks later, it announced the layoff of approximately 6,000 workers, its largest reduction since 2023, cutting nearly three percent of its global workforce, per the Associated Press. The company’s AI business simultaneously reported an annualized revenue run rate of $37 billion, growing at 123 percent year over year, per Statista. The operational logic was explicit: Microsoft was not cutting because it was failing. It was cutting because AI had allowed it to define success with fewer people, and the stock market would reward the restructured cost structure regardless of what happened to the workers who exited it.
The Capital Expenditure Race and What $700 Billion Does Not Guarantee
The AI infrastructure buildout currently underway is not a normal capital investment cycle. Its scale makes the comparison instructive. In 2025, Microsoft, Alphabet, Meta, and Amazon committed a combined $300 billion or more in capital expenditures to AI infrastructure, per analysis compiled by ValueAddVC from company earnings reports. In 2026, those commitments have been revised sharply upward: Amazon to $200 billion, Alphabet to between $91 and $93 billion, Meta to between $125 and $145 billion, Microsoft to $190 billion for its fiscal year ending June 2026, per CNBC and Fast Company reporting from late April and early May 2026. Combined 2026 expenditures by those four companies approach $700 billion. Bank of America analysts, tracking the trajectory, project aggregate AI capital spending by the five primary hyperscalers surpassing $1 trillion annually by 2027. McKinsey separately estimates that $7 trillion in data center investment will be required by 2030 to meet projected AI demand.
This capital is being deployed into GPU clusters, custom silicon, and data centers. What it is not being deployed into is the workforce or the consumer economy that provides the revenue base for the products this infrastructure is intended to generate. The distinction matters enormously.
The free cash flow consequences of this spending are already visible. Pivotal Research projected that Alphabet’s free cash flow would fall roughly ninety percent in 2026, from $73.3 billion in 2025 to $8.2 billion, per CNBC’s February 2026 reporting. Meta’s free cash flow in Q1 2026 was $1.2 billion, down from $26 billion in the same quarter of the prior year. Bank of America credit strategists, analyzing capital spending trends, noted that AI capex was tracking toward ninety-four percent of the hyperscalers’ combined operating cash flows net of dividends and buybacks for 2025 and 2026. These companies are financing the gap between their operational cash generation and their capital commitments through corporate debt and off-balance-sheet vehicles.
When Mark Zuckerberg was asked by an analyst on Meta’s April 2026 earnings call to specify the indicators he uses to assess whether the company is on a path toward return on investment in AI, his answer was: “That’s a very technical question.” He then described building products for billions of users and monetizing them at scale. The formula he described is the same formula Meta has used for two decades. It contains no mechanism specific to justifying $125 to $145 billion in annual capital spending. The analyst’s question went unanswered. Meta stock fell on the call, then recovered partially.
The market’s tolerance for deferred proof of returns on AI investment has been the defining feature of the current bull run. That tolerance rests on one underlying assumption: that the AI productivity gains companies are promising will eventually materialize at sufficient scale to justify the infrastructure being built. A February 2026 National Bureau of Economic Research study surveying 6,000 executives found that the vast majority reported no productivity impact from AI whatsoever. Aggregate U.S. productivity data told a different story, with output per worker nearly doubling its decade-long average in 2025. The contradiction sits at the center of current uncertainty: something is producing productivity gains in the aggregate data, but most of the companies deploying AI cannot feel it in their operations.
This dynamic has a name. Thousands of CEOs admitting AI had “no impact on employment or productivity,” as Fortune reported in February 2026, was described by economists as a resurrection of Solow’s Paradox: the observation by Robert Solow in 1987 that computers were visible everywhere except in the productivity statistics. The paradox resolved in the late 1990s, a decade after the computing infrastructure was built, when organizational restructuring finally caught up to the technology. Whether the AI version resolves on a similar timeline, or faster, or not at all, is the question the $700 billion annual spending bet is being placed on without a definitive answer.
The Labor Market Below the Surface
The information sector, which includes software, telecommunications, data processing, and computing infrastructure services, shed 342,000 jobs between its peak in November 2022 and April 2026, a decline of 11.0 percent, per the Bureau of Labor Statistics employment situation report released May 8, 2026. The BLS noted specifically that “information employment is down by 342,000, or 11.0 percent, since its most recent peak in November 2022.” November 2022 is two months after ChatGPT’s training data was cut off. It is the precise moment generative AI tools began proliferating into the professional economy.
The correlation is not conclusive, and economists have cautioned against treating it as proof of structural AI-driven displacement. Broader economic factors, rate increases, post-pandemic correction, all played roles. But the timing and sector specificity are not coincidental, and the data on specific role categories makes the directional argument more legible. Writing job postings have fallen thirty percent since 2022. Software and web development roles dropped twenty-one percent. Engineering positions declined ten percent. Freelancers in writing roles saw an average two percent monthly job decline and a five percent monthly earnings drop after generative AI tools became widespread, according to Cornell University research cited in an ALM Corp analysis published March 2026.
Employment among workers aged twenty-two to twenty-five in software development occupations has fallen twenty percent from its late 2022 peak, per The World Data analysis compiling BLS and Goldman Sachs data. This is the entry-level cohort: the people who would have taken junior roles at technology companies, learned their craft, and become the productive mid-career professionals who drive the technology industry’s output a decade from now. A twenty percent decline in entry-level employment in software is not a temporary correction. It is a structural compression of the professional pipeline.
The broader labor market context from May 2026 data amplifies this. Deloitte’s March 2026 U.S. economic forecast noted that average monthly nonfarm payroll gains had been just 14,000 during the six months through January 2026, far below the average of 122,000 in 2024. The Conference Board’s April 2026 U.S. Leading Economic Index report, released in May, stated that the index had “pulled back sharply in March,” as building permits declined and consumer expectations weakened. The Conference Board revised its full-year 2026 GDP forecast to “well below 2 percent,” settling at 1.6 percent year over year. The Philadelphia Federal Reserve’s first-quarter 2026 Survey of Professional Forecasters projected job gains averaging just 48,500 per month in 2026, against a monthly average of 85,000 in 2024 private-sector employment gains and 122,000 total.
The CEPR’s Q1 2026 GDP preview made the composition problem plain. Q1 2026 GDP came in at an annualized 2.0 percent, up from Q4 2025’s revised 0.5 percent. But CEPR’s Dean Baker noted that most areas of consumer spending showed “little or no growth.” The GDP expansion was carried by government spending rebounding after the Q4 2025 federal shutdown, by war-related defense expenditures, and by data center construction and equipment investment related to AI. Consumer goods spending was “close to flat” in Q1 2026. Business investment in data centers is genuine investment. It does not translate into broad consumer income. The workers building data centers in rural Virginia and suburban Texas earn construction wages. The workers running the AI compute clusters within those facilities are a small, highly specialized, well-compensated group. Neither category constitutes the mass consumer income base that sustains a $29 trillion economy.
The Fractured Consumer Base and Who Is Actually Spending
The Federal Reserve Bank of Boston’s August 2025 analysis of large-scale credit card data found that since 2022, real aggregate spending, adjusted for inflation, had been propelled almost entirely by the highest-income consumers. Spending adjusted for inflation among the bottom eighty percent had stagnated. A KPMG analysis of Federal Reserve data from November 2025 made the divergence concrete: the top twenty percent of households now account for nearly two-thirds of all consumption. The top 3.3 percent of households by income have increased spending the most. For everyone below the sixty-seventh percentile by income, real spending has been flat.
This is not a marginal observation. It describes a consumer economy that is effectively being powered by one income quintile, the members of which own the overwhelming majority of the stock market wealth that is appreciating. Their spending is driven not primarily by wages but by asset appreciation. The Boston Fed made the feedback loop explicit: “a red flag would appear if the stock market declines, given how wealth effects stimulate high-end consumption.” The spending resilience that keeps GDP positive is, at its foundation, collateralized by stock prices. The AI trade, in other words, is the consumer base.
The credit data exposes the other side. According to The Century Foundation’s March 2026 analysis of the University of California Consumer Credit Panel, approximately 68 million Americans, nearly one in three cardholders, carried credit card utilization above thirty percent as of Q4 2025, the threshold above which persistent debt and reduced credit health become probable. These cardholders collectively hold approximately $800 billion in outstanding credit card balances. More than 27 million of them, over one in nine cardholders, cannot afford to pay more than the minimum payment due each month. At average outstanding balances, a minimum-payment cardholder in this group pays approximately $251 per month while accruing interest on the remaining ninety-eight percent of their balance. Experian’s 2025 consumer credit review noted that average credit card APRs remained “well above 20 percent for most consumers” despite four Federal Reserve rate cuts in the preceding eighteen months.
A KPMG survey cited in the March 2025 consumer credit report found that thirty-three percent of Americans carry more credit card debt than savings. Separately, the number of Americans who could make only the minimum payment on their credit card grew by 6.3 million between early 2018 and the end of 2025, a 29.6 percent increase in the population of Americans trapped in compounding consumer debt.
The earnings-per-share records, the buyback records, the stock market records, rest on this foundation. The corporate profit margin expansion was achieved by reducing the wages of the same households whose consumption those corporations depend upon. The compression is structural. It cannot resolve itself without an external intervention that is not currently visible in any policy framework being advanced in Washington.
The AI Layoff Trap: Why Firms Cannot Stop
The theoretical structure of what is happening was articulated formally in a March 2026 working paper by Brett Hemenway Falk and Gerry Tsoukalas at the University of Pennsylvania. Their paper, “The AI Layoff Trap,” published through arXiv, models the following mechanism. When AI displaces a human worker, that worker is also a consumer. The displaced worker’s lost income reduces demand across the economy. When one firm automates to reduce costs, its competitors face a choice: automate at the same rate or lose market share. The competitive structure forces each firm to automate regardless of the aggregate economic consequences, because those consequences fall outside any individual firm’s balance sheet. Rational individual behavior produces collectively irrational outcomes. The paper finds that standard policy remedies, including wage adjustments, retraining programs, worker equity participation, and universal basic income, cannot resolve the trap in a competitive market. The only intervention that changes the incentive structure is a Pigouvian automation tax, a direct cost placed on the labor displacement externality.
Their argument is theoretical. The empirical trace of the trap is visible in the labor data from every sector where AI adoption is measurable. What is happening across the information sector, writing markets, junior software roles, call centers, and data entry functions is not an accident of timing. These are the roles most directly substitutable by generative AI, and they are declining in the precise sequence the substitution thesis predicts.
Oxford Economics, reviewing the aggregate macroeconomic evidence in January 2026, pushed back. Their analysis argued that firms do not appear to be replacing workers with AI on a significant scale, and that the productivity data does not show the output-per-remaining-worker surge that genuine mass substitution would produce. They estimated that AI-attributed layoffs represented 4.5 percent of total reported job losses in 2025. The broader job loss total was driven by standard market and economic conditions, DOGE cuts, tariff disruptions, and retailer contractions. Oxford’s framing is accurate on the aggregate data. What it does not resolve is the sectoral and demographic specificity. The aggregate labor market contains 160 million workers. A structural collapse in information employment, in junior technology roles, in knowledge-work entry levels, does not show up as a recession signal in the broad headline. It shows up as a slow, quiet compression of the income floor for the most AI-exposed workers, who happen to be concentrated in the age brackets that would otherwise form the next generation of the consumer class.
The IMF’s Staff Discussion Note estimated that forty to sixty percent of jobs in advanced economies are exposed to AI. Goldman Sachs research estimated that approximately sixty-three percent of U.S. work hours are exposed to AI, with twenty-five to fifty percent of tasks directly automatable. What “exposed” means at scale, and over what timeline, is the question. Cognizant’s “New Work, New World 2026” report estimated that ninety-three percent of U.S. jobs can be partially performed by AI, and that companies could unlock over $4.5 trillion in labor productivity through AI solutions. That $4.5 trillion in productivity extraction is, in economic terms, $4.5 trillion in labor cost reduction. Reduced labor costs mean reduced wages flowing to the consumer economy. The question the AI investment thesis does not answer is where the consumer demand comes from when the productivity gains have been extracted.
The Political Accelerant
The AI-driven workforce compression is not occurring in a stable policy environment. It is occurring inside a political climate that has added compounding stress to an already fragile consumer base.
DOGE, the executive initiative led by Elon Musk and given statutory backing, oversaw the elimination of more than 62,530 federal jobs in the first two months of 2025, a 41,311 percent increase in federal layoffs year over year, per Human Resources Director tracking data cited by the Chicago Sun-Times. By year’s end, DOGE-linked cuts had reached nearly 300,000 federal positions per Challenger, Gray and Christmas data, with 63,583 additional private and nonprofit sector layoffs directly attributed to cancelled federal contracts. The April 2026 BLS employment situation report confirmed that federal government employment continued to decline by 9,000 positions in April alone. The federal government is the single largest employer in the United States. Its systematic reduction is not a neutral efficiency exercise. It is a demand shock in the specific regional and demographic labor markets where federal employment is concentrated.
The tariff regime added a supply-side inflation layer to the demand compression. The average effective tariff rate on goods imports ended 2025 at 12.5 percent, against a 3 percent pre-administration baseline, per the Center for American Progress analysis by economist Ernie Tedeschi. For households in the bottom eighty percent of income, who spend a larger share of their incomes on goods than services, tariff-driven price increases constitute a direct reduction in real purchasing power at the same time their labor market position has weakened. Q4 2025 GDP came in at a revised 0.5 percent annualized, the weakest quarter since the pandemic. The Conference Board’s LEI declined sharply in March 2026 even as Q1 2026 recovered to 2.0 percent, driven by government shutdown reversal and data center investment rather than consumer demand.
The Iran war, launched in late February 2026, added energy price volatility. Crude oil is trading near $100 per barrel as of May 2026 per Trading Economics data. Real average hourly earnings grew just 0.3 percent year over year through April 2026 after adjusting for inflation, per BLS. Nominal wages rose 3.6 percent annually. The gap between nominal and real wage growth, approximately 3.3 percentage points, is energy-driven inflation eating into paychecks while labor markets show “little net change” across twelve months.
The Federal Reserve, facing this combination of weak demand and energy-driven inflation, held rates at the 3.5 to 3.75 percent range at its April 29 meeting, with four dissents, the most since 1992. The CME FedWatch Tool showed markets pricing a seventy percent probability of rates remaining unchanged through year’s end. Kevin Warsh, taking over as Fed chair in May 2026, inherits an economy where the standard tools for managing either inflation or stagnation produce worse outcomes for the other. The Conference Board put it plainly: “higher oil prices and supply chain tensions will likely place additional upward pressure on inflation and further reduce consumers’ purchasing power.”
This is the definition of stagflation at the structural level. Not the 1970s variety, when supply shocks hit an economy at full capacity and wages were indexed to inflation. The 2026 variety: technology-driven displacement reducing labor’s bargaining power while energy and goods inflation erodes what purchasing power remains, in an environment where monetary policy cannot address both simultaneously and fiscal policy is being used to shrink the government workforce rather than support the consumer base.
The Profit-Demand Contradiction at the Center of the Machine
The analytical thread running through all of these data sets converges on a structural problem that American corporations have deferred, but cannot permanently avoid.
The corporate profit margin record of 14.0 percent requires sustained revenue. Revenue at scale requires consumer demand. Consumer demand requires household income. Household income in a wage economy requires employment at sufficient levels and wage rates. The AI investment thesis accelerates the substitution of employment with capital, which reduces the wage bill, which expands margins in the short term, which drives stock prices, which enriches the top decile, which sustains consumption for the top quintile, which keeps GDP positive, which lets the cycle continue. The mechanism is coherent within its own terms. What it cannot solve is the erosion of its own base.
The $44 trillion in stock market wealth held by the top ten percent is the collateral for the consumption that sustains two-thirds of GDP. That wealth is concentrated in approximately eight companies at the S&P 500’s apex, whose valuations require AI to deliver productivity gains that most of their own C-suites cannot yet verify in their operations. The bottom eighty percent of American households, stagnant on real consumption, are servicing $800 billion in high-interest credit card debt at rates above twenty percent. Writing jobs down thirty percent. Software entry-level roles down twenty-one percent. Information employment down eleven percent from its 2022 peak. The seasonal retail workforce shrinking. The federal payroll declining every month.
None of these figures individually constitutes a crisis. Collectively they describe a consumer economy that has been slowly hollowed from below while its financial instruments have been inflated from above. The S&P 500 at 7,500 is an accurate measurement of corporate ownership value at the current moment. Whether it is an accurate measurement of corporate ownership value eighteen months from now depends entirely on whether the AI productivity gains being financed by $700 billion in annual capital spending can be monetized through a consumer economy that the same companies building those gains are simultaneously dismantling.
The Ford logic, whatever its limitations, contained an internal coherence that the current corporate calculus lacks. When Henry Ford doubled line workers’ wages to $5 a day in 1914, his rationale, stated without apology, was that workers who could not afford the product were a production problem. The AI investment thesis contains no equivalent acknowledgment. It proceeds as though the consumer base is a natural resource that requires no maintenance, and whose depletion will not affect the corporations extracting value from it.
Nobody on Wall Street has a financial model for what happens when that assumption fails. The Q1 2026 earnings season, with 84 percent of companies beating estimates at the highest surprise rate since Q2 2021, suggests it has not yet. The April jobs report, with 803,000 Americans moving into fresh labor distress in a single month while payrolls show “little net change over the prior 12 months,” suggests the timeline is shorter than the current market multiple implies.
A Note Toward the Next Investigation
There is a dimension to this structural problem that the U.S. frame cannot fully contain. What is playing out on American balance sheets at the premium end of the income distribution is being replicated, at the bottom end, across the entire Global South.
The Industrial Revolution did not distribute its gains or its disruptions evenly. The societies that built the machines captured the productivity premium. The societies whose raw materials and labor fed those machines absorbed the displacement without the institutional frameworks to manage it. Britain industrialized over roughly a century. India absorbed the consequences of that industrialization over several centuries that are not finished. The adjustment period was measured in generations, and in some cases it is still not complete.
AI is following the same structural geography. The $700 billion being deployed in 2026 is being deployed in the United States, in data centers in Texas and Virginia, on chips designed in California and manufactured in Taiwan. The productivity gains accrue to companies listed on the S&P 500 and owned, eighty-seven percent, by the wealthiest ten percent of American households. The countries of the Global South face a compounding disadvantage: the call centers of Manila, Bangalore, Nairobi, and Cairo that absorbed the last wave of labor displacement from prior automation are precisely the sectors generative AI is targeting now. The content moderation roles that connected workers in Kenya and the Philippines to global technology supply chains, the data labeling jobs that provided digital economy income across Sub-Saharan Africa, the back-office services that allowed India’s middle class to form, are the entry-level AI-disrupted positions whose elimination Oxford Economics is currently studying in the American data.
For the developed world, AI is a disruptive transition with adjustment mechanisms, imperfect and strained, but present. For the underdeveloped world, it is a second structural displacement arriving before the first one fully resolved. The Industrial Revolution took the developing world decades to adapt to, and in many cases the adaptation is still incomplete. AI will not wait decades for institutional frameworks to form. Its deployment cycle is measured in product release cycles.
That investigation will come separately. But no accounting of what the S&P 500 at 7,500 means in May 2026 is complete without noting that the question nobody in corporate America has answered, “who buys the product when the consumer has been automated out of their income,” has one answer in the United States and a catastrophically different answer in every economy where displacement is structural, safety nets are absent, and the AI infrastructure generating these returns is owned by no one within their borders.