Owning Beats Working: AI Wealth and the American Dream on Credit

Two Engines: Data Centres and the Wealthiest Tenth

The American economy is building data centres and not much else. Amazon, Microsoft, Alphabet and Meta plan to spend roughly 725 billion dollars on capital expenditure in 2026, an increase of about 77 percent on the 410 billion they spent the year before, and the overwhelming share of it goes into server halls, graphics processors, custom silicon and the electricity to run them. Goldman Sachs expects the four of them to commit 5.3 trillion dollars between 2025 and 2030.

Set that against what the same economy did for the people living in it. Business investment in equipment rose 15.2 percent in the second quarter while investment in structures fell for the tenth consecutive quarter, which means the cranes over office blocks and warehouses have been coming down for two and a half years. In June the whole American economy produced 57,000 new jobs. The largest corporate building programme in recorded history is under way, and it employs almost nobody.

The money reaching households comes from the same source by a different route. Consumer spending accelerated to 3.2 percent in the second quarter, carried by the upper half of the income distribution, where the wealthiest tenth of American households now accounts for close to half of all consumer spending in the country. Michele Raneri of TransUnion, reviewing the first quarter, described the K-shaped economy as alive and well, which is an unusually direct way for a credit bureau to say that one country is having two different years.

What 2008 Made Visible: The Recovery That Went One Way

The shape has a longer history than the rally that made it visible. In 2007 the median American family kept two thirds of everything it owned in the house it lived in, while the wealthiest one percent kept roughly nine tenths of its assets in shares, securities and company holdings. When the crash came, house prices fell twenty-three percent and share prices fell twenty-one, and for a few months the losses looked evenly distributed. Then the recovery arrived and separated them. Shares came back within a few years. Housing took the better part of a decade, and in a great many American counties it never came back at all.

What followed is the most thoroughly documented economic experiment of the century. Between 2009 and 2012 the incomes of the top one percent rose 31.4 percent while the bottom ninety-nine percent gained four tenths of one percent, which means the top one percent absorbed ninety-five percent of everything the recovery produced. Emmanuel Saez, who assembled those figures from tax records, found six years on that the bottom ninety-nine percent had recovered around sixty percent of what they lost. The proposition that wealth accumulating at the top eventually reaches the base was given fifteen years and the largest monetary expansion in modern history to demonstrate itself, and the record of what it produced is now public.

The Dream on Credit: Fifteen Years of Believing

None of this ended the American Dream. It made the structure of it visible in the data while leaving the belief entirely intact, and the belief went on doing the work. The middle kept buying through the 2010s on credit, on rising house prices that later stopped rising, and on the assumption that the next decade would be better than the one before.

Aspirational luxury is the material form of that assumption. A broad base, a narrow summit, and a route between them that a person can believe she is walking, with the object in the bag serving as the receipt for the climb. The pyramid stood on expectation long after it stopped standing on wealth.

Owning and Working: Why the Rally Cannot Reach the Middle

In 2026 the expectation is meeting the arithmetic. The median American household earned 83,730 dollars in 2024, a figure the Census Bureau itself declined to call a significant change from the year before, and nothing since has moved it in a way a family would notice. Typical house values have risen by around half since 2019. A median mortgage now takes 29.1 percent of the median household income before a dollar of tax is paid, which is a way of saying that owning has become the thing that makes people wealthier while working has become the thing that keeps them level.

At the bottom of the wage scale the movement went the other way, with real pay at the tenth percentile falling three tenths of a percent to 14.56 dollars an hour in 2025. American households added 736,000 millionaires in the same period, most of them carried there by an equity market that artificial intelligence had lifted to records, and the wealthiest tenth of American households owns close to nine in every ten dollars of stock. A rally in shares reaches a household that owns shares.

Torsten Slok of Apollo has spent the year calling the resulting concentration a single point of failure, and translated into purchasing power his point becomes sharper still, because the same tenth of households that owns the shares also carries half of the consumption. The revenues of the companies this capital is being spent on behalf of, OpenAI and Anthropic among them, remain a fraction of the infrastructure going up for them. Microsoft is carrying an Azure order backlog of around 80 billion dollars it cannot fill, because the constraint is electricity.

The Twelve-Room Economy: Who Closes First

The businesses that register this first are the small ones, and the evidence for that arrived this summer. An estimated 8,171 restaurants closed across the United States and Canada in the first half of 2026, and while chains and independents split the overall total almost evenly, the composition tells the real story. Independent operators accounted for 2,318 of the 2,793 closures in casual and family dining, which is 83 percent of the closures in exactly the segment where a family sits down to eat.

In Washington the pattern shows up in traffic rather than closures. Seventy-six percent of mid-priced restaurants reported fewer diners in 2025, against fifty-six percent of fine dining establishments, and no casual dining restaurant in the city reported sales growth at all while twenty-two percent of fine dining did. Shawn Townsend, who runs the city’s restaurant association, put it plainly when he said that the restaurants middle-class families have long depended on are disappearing.

Hotels show the same division with the same clarity. Luxury and upper-upscale properties in the United States are running at 67.3 percent occupancy and an average daily rate of 281 dollars, while midscale and economy hotels sit at 54.5 percent and 86 dollars, and the analysts describing this call it a barbell market with the weight concentrated at the top. Independent hotels, measured across ninety million bookings worldwide, lost 5.4 percent of revenue per available room over the same period in which branded luxury grew.

A group with a portfolio responds to this by moving upmarket, raising rates at the top and reporting growth for the year. A house with twelve rooms cannot follow, because its guest is a person with a decent income and a considered way of spending it, someone who reads for an hour before booking and pays more for a place that keeps a cook, a garden and a supplier down the road. That guest is the middle, and when the mortgage and the insurance and the weekly shop take more of the month, the three nights in the valley are what goes. The restaurant that buys from one farm and the workshop that repairs what it once sold depend on the same person, and none of them has a second clientele waiting.

What follows travels further than the individual closure. When the restaurant goes, the farm loses the account that justified the way it grows. When the small hotel gives up, the reason to drive to that valley goes with it, and the shop on the road down loses the traffic within two seasons. Regional economies hold together through a small number of businesses that pay properly and buy locally, and those businesses run on a customer who is currently being removed from the equation.

Cabanis Was Right: Reading the Record Backwards

Cécile Cabanis, the chief financial officer of LVMH, described the mechanism on the results call in the plainest possible terms when she said that wherever there is wealth creation, there is a strong appetite for luxury. She is right, and the sentence carries further than she intended it to. The American luxury market is posting record numbers in 2026 because wealth is being created in America at a speed and a concentration without recent precedent, and those numbers are being reported as evidence that the economy is healthy. Read from the other end, they are a measurement of how far one layer has travelled from the rest.