Dimitri Berishvili

“Is it a little bit homophobic to focus on the straights of Hormuz rather than the gays of Hormuz?” No Kings protester, completely serious: “Yes, absolutely, I agree.”

I just published WHO IS LEFT TO BUY? medium.com/p/who-is-left-…

WHO IS LEFT TO BUY? The Economic Contradiction at the Heart of AI Replacement Fears Everyone is talking about AI replacing humans. Replacing jobs, replacing industries, replacing entire economic sectors. The conversation has turned into a kind of panic, as if one morning we will wake up and every role from truck driver to lawyer to software engineer will be gone, swallowed by machines that work faster and cheaper and never ask for a raise. I am not here to tell you AI is not powerful. It is. I am not here to tell you jobs will not change. They will. But the fear that AI will replace all humans, take all the money, and leave nothing behind misses something so fundamental that it is almost embarrassing that more people are not saying it. If AI replaces every worker, there is no one left to buy anything. That is not a philosophical musing. That is the entire structure of the modern economy. The companies being valued at trillions of dollars on the promise that AI will generate limitless efficiency are the same companies whose revenue depends on humans having money to spend. You cannot have it both ways. You cannot celebrate the elimination of the workforce and simultaneously expect the marketplace to keep functioning. The math does not work. The economics do not work. And if you follow the logic of total replacement to its endpoint, the whole thing collapses under its own weight. The Loop That Runs Everything Before getting into theory or history, start with the most basic model in economics. It is called the circular flow of income and it is taught in the first week of every introductory econ class on the planet. The concept is simple. Businesses produce goods and services. To do that, they hire workers. Workers get paid wages. Workers take those wages and spend them on goods and services. That spending becomes revenue for businesses. Businesses use that revenue to hire more workers, produce more, and the cycle continues. This is not some abstract theory. It is how the real economy actually works, and the numbers prove it. According to the Federal Reserve (FRED data series DPCERE1Q156NBEA) and the Bureau of Economic Analysis, Personal Consumption Expenditures hit $19.667 trillion in Q4 2025. That is 68% of the entire U.S. GDP. Not a third. Not half. Two thirds. And that 68% is the highest share in modern history, well above the long-term average of 64.4%. Let that number settle. 68%. That means consumer spending is not a piece of the economy. It is not a contributing factor. It IS the economy. Everything else, government spending, business investment, exports, all of it together accounts for the remaining 32%. When politicians talk about economic growth, when CEOs report earnings, when analysts forecast GDP, what they are really measuring, the vast majority of the time, is whether or not regular people are spending money. Now run the thought experiment. AI replaces all workers. Every single one. Who earns the wages? Nobody. Who spends the wages? Nobody. Where does the $19.667 trillion in consumer spending come from? Nowhere. What happens to the 68% of GDP that depends on that spending? It evaporates. The loop is not just disrupted. It is destroyed. The economy does not slow down. It ceases to function in any recognizable form. How Are You Going to Get Robots to Buy Fords? This argument is not new. It was made seven decades ago, and it was made perfectly. In the early 1950s, the United Auto Workers union leader Walter Reuther was touring a newly automated Ford engine plant. The machines were impressive, rows of mechanical arms doing the work of hundreds of men. A company official, feeling smug about the display of technological progress, turned to Reuther and asked: “How are you going to collect union dues from these guys?” Reuther did not miss a beat: “How are you going to get them to buy Fords?” The earliest documented version of this exchange appeared in a UAW-CIO conference report from 1955. The specific wording varies across retellings, but the logic is airtight and it has never been answered. Not in 1955. Not now. Henry Ford understood a version of this even earlier. In 1914, he raised his workers’ wages to $5 a day, more than double the going rate. The popular story is that Ford was being generous. He was not. He was being strategic. Ford recognized that if the people building his cars could not afford to buy them, there would never be a mass market for automobiles. As Ford himself put it: “Country-wide high wages spell country-wide prosperity, provided the higher wages are paid for higher production.” The principle has not changed. Production without purchasing power is inventory, not commerce. You can build the most efficient AI system in the world, capable of producing goods and services at a fraction of the cost, but if nobody can afford to buy what it produces, you have built the most impressive warehouse in history. Nothing more. Keynes Saw It Coming, but Not Like This John Maynard Keynes is one of the most influential economists who ever lived. In 1930, he wrote an essay called “Economic Possibilities for our Grandchildren” where he coined the term “technological unemployment”, defining it as “unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.” Keynes saw the problem clearly. Technology would eliminate jobs faster than the economy could create new ones. But he thought this was temporary, a phase of adjustment. He predicted that by 2030, the standard of living would be four to eight times higher and people would only need to work 15 hours a week. The first prediction was roughly correct. The second was wildly wrong, because consumer demand kept expanding as new products were invented and marketed. People did not work less. They just bought more. But here is the critical point for this argument: Keynes assumed workers would still exist. His entire economic framework, what we now call Keynesian economics, is built on the concept of aggregate demand, the total spending in an economy. Aggregate demand is what drives hiring, production, investment, and growth. And the largest single source of aggregate demand is consumer spending, which comes from wages, which come from employment. Pull the workers out of the system and the Keynesian framework does not just weaken. It collapses. No employment means no wages. No wages means no consumer spending. No consumer spending means no aggregate demand. No aggregate demand means no economic activity. It does not matter how productive your AI is if there is no demand for what it produces. Keynes understood that demand drives supply, not the other way around. Remove the demand and supply becomes irrelevant. We Are Not Horses In 1983, Nobel Prize winning economist Wassily Leontief made an analogy that gets cited constantly in AI discussions. He compared human workers to horses. In 1910, horses were everywhere. Farming, transportation, war, they were essential to the functioning of society. Then the internal combustion engine came along. By 1960, the horse population in the United States had fallen by 85%. Horses were not retrained. They were not given new careers. They were not helped through a “transition period.” They became economically useless and the economy moved on without them. Leontief wrote: “The role of humans as the most important factor of production is bound to diminish in the same way that the role of horses was first diminished and then eliminated.” — Wassily Leontief This quote gets used to scare people and it is effective. But the analogy has a fatal flaw, and it is the flaw that supports this entire essay. Horses were never consumers. A horse did not earn a wage. A horse did not pay rent. A horse did not buy groceries, subscribe to streaming services, take vacations, or finance a car. Horses existed purely on the production side of the economy. When technology replaced them, the economy lost labor capacity but it did not lose demand. The people who previously relied on horses simply switched to machines and the consumer economy continued. Humans are fundamentally different. We are both producers and consumers. We sit on both sides of the economic equation simultaneously. When you replace a horse, you lose a beast of burden and nothing else changes on the demand side. When you replace a human worker, you lose a producer and a customer. The economy takes the hit from both directions at the same time. Production capacity might stay the same or even increase, but purchasing power drops. The system is not just losing an input. It is losing the output’s only reason for existing. Marx Called It 150 Years Ago You do not have to be a Marxist to recognize that Karl Marx identified this exact structural problem over 150 years ago. In Chapter 25 of Capital, “The General Law of Capitalist Accumulation,” Marx described what happens as businesses invest more in machinery (what he called constant capital) and less in workers (variable capital). The pattern he identified is straightforward. As technology improves, businesses replace workers with machines because machines are cheaper per unit of output. This creates what Marx called an “industrial reserve army” of unemployed workers. These workers, no longer earning wages, can no longer buy the goods being produced. Production capacity expands while the market for that production shrinks. The system produces more than it can sell. This is the underconsumption problem, and it has been observed in various forms throughout economic history. Every major economic crisis has some version of it: too much supply chasing too little demand. The Great Depression. The 2008 financial crisis. Each time, the mechanism was different, but the underlying math was the same. People who do not have money do not spend money. When enough people stop spending, the system breaks. Full AI replacement is this dynamic accelerated to its absolute extreme. It is not a gradual shift in the composition of capital. It is the total elimination of variable capital. Every dollar of income flows to the owners of the machines and zero flows to workers because there are no workers. Marx predicted that capitalism would produce its own crises through this mechanism. He could not have imagined a technology capable of making the contradiction this stark. The MIT Evidence: Four Decades of Proof Everything above is theory. Important theory, but theory. Now here is the data. Daron Acemoglu is an Institute Professor at MIT and one of the most cited economists in the world on the topic of technology and labor. Along with his co-author Pascual Restrepo, he built what is called the task-based framework for understanding how automation affects the economy. Their research, published through the National Bureau of Economic Research (NBER Working Paper 24196), breaks it down into two competing forces. First, there is the displacement effect. When machines or AI take over tasks that humans used to do, demand for human labor drops. Wages come under pressure. Workers either accept less or find themselves unemployed. This is the part everyone talks about. Second, there is the productivity effect. When automation makes production cheaper, the savings can theoretically increase demand for labor in other areas. New tasks get created. New industries emerge. This is the part optimists point to. Here is what Acemoglu’s four decades of data actually show: the displacement effect has been winning. Automation has raised productivity. Corporate profits have multiplied. But wages have stagnated. The gains went to capital, not to labor. Writing in the IMF’s Finance and Development magazine, Acemoglu and Simon Johnson stated it directly: “Wages are unlikely to rise when workers cannot push for their share of productivity growth. Artificial intelligence may boost average productivity, but it also may replace many workers while degrading job quality for those who remain employed.” — Acemoglu & Johnson, IMF Finance & Development This is the consumer paradox playing out in slow motion right now, before AI has even reached its full potential. If workers do not share in the productivity gains, they cannot sustain the demand that the economy needs. The evidence is not hypothetical. It is forty years deep. The Inequality Death Spiral Joseph Stiglitz, Nobel laureate and professor at Columbia University, takes this one step further. His body of work demonstrates that inequality is not just a social or political problem. It is an economic problem. Specifically, it is a demand problem. The mechanism is simple. A middle class family that earns $80,000 a year spends most of it. Housing, food, transportation, healthcare, education, entertainment, the money circulates. A billionaire who earns another $80 million does not spend proportionally more. They might buy another property or another asset, but assets are not consumer spending. The money pools at the top instead of cycling through the economy. Stiglitz’s paper “Inequality and Economic Growth” published through Columbia Business School makes the case that rising inequality directly suppresses aggregate demand because wealthy individuals spend a smaller share of their income. The math is unforgiving: when income concentrates at the top, less of it gets spent on goods and services, which means less revenue for businesses, which means fewer jobs, which means even less spending. It is a spiral. Now push this to the extreme of full AI replacement. If all income flows to the owners of AI systems and zero flows to workers (because there are no workers), you have the most extreme version of inequality possible. It is not a wealth gap. It is a wealth cliff. A small group of people own everything and the market for their products has vanished. Stiglitz has warned directly about AI’s potential to create this scenario, telling Scientific American that AI “may be an ally of the employer and weaken workers’ bargaining power even more, and that could increase inequality even more.” You cannot build a consumer economy without consumers. You cannot sustain trillion dollar valuations selling to a customer base of zero. The inequality argument is not about fairness. It is about arithmetic. The Trillion Dollar Question Nobody Is Asking Let us talk about the actual numbers driving the current AI boom, because they tell a story that contradicts the replacement narrative. Between 2023 and 2025, the hyperscalers, Microsoft, Alphabet, Meta, Amazon, collectively poured over $400 billion into AI infrastructure. Mostly GPU clusters and data center construction. Deutsche Bank projects cumulative spending of $4 trillion on AI data centers through 2030. These are not small bets. These are the largest capital expenditures in the history of technology. And the returns? In June 2024, Sequoia Capital’s David Cahn identified a $600 billion gap between the revenue required to justify the AI infrastructure buildout and the actual earnings of the AI ecosystem. By early 2026, that gap has not closed. It has widened. The PwC 2026 Global CEO Survey, covering 4,454 CEOs across 95 countries, found that 56% say they have gotten nothing out of their AI investments. Not “less than expected.” Nothing. Only 12% reported that AI both grew revenues and reduced costs. A parallel NBER study of 6,000 executives across the U.S., U.K., Germany, and Australia found firms forecasting just a 0.7% employment cut over the next three years. Not the mass displacement everyone fears. 0.7%. Harvard economist Jason Furman ran the numbers on where U.S. GDP growth was actually coming from in 2025. His finding: strip out data center construction and GDP growth in the first half of 2025 was 0.1%. Essentially zero. The AI boom is currently being sustained by companies spending money on building AI, not by AI generating value that consumers are paying for. And data centers, once built, employ almost nobody compared to factories or office campuses. The money goes in but very little comes back through the wage-spending loop that drives 68% of the economy. Here is the contradiction that nobody in Silicon Valley wants to talk about. These companies are valued as if AI will replace everything and generate infinite efficiency. But if AI actually replaced all workers, the consumer base generating 68% of GDP disappears. The companies would be worth trillions on paper with nobody to sell to. Their valuations depend on a world where AI boosts productivity but consumers still have income to spend. Those two things are in direct tension, and the more AI replaces, the worse the tension gets. Every CEO Is Rational. The System Is Not. Keynesian economics has a concept called the Paradox of Thrift. It goes like this: it is rational for any individual person to save more money during uncertain times. But if everyone saves at once, total spending drops, businesses lose revenue, they lay off workers, those workers spend even less, and the economy shrinks. What is rational for the individual is destructive for the collective. Full AI replacement is this paradox on a completely different scale. It is perfectly rational for any single CEO to look at their cost structure and say: I can replace these 500 employees with AI and save $50 million a year. That CEO is making a smart business decision. Margins improve. Shareholders are happy. The board applauds. But if every CEO in every company across every industry makes the same decision at the same time, they collectively destroy the customer base that all of them depend on. Those 500 employees were not just costs. They were customers of other businesses. They bought groceries, paid rent, subscribed to services, took trips, bought insurance, financed cars. Multiply that across millions of workers and you have a demand collapse that no amount of AI efficiency can fix. This is what economists call a coordination failure. Each individual actor behaves rationally, but the collective result is catastrophic. The individual incentive to automate is real. The systemic consequence of universal automation is economic destruction. These two facts coexist, and the second one makes the first one self-limiting. The Brake the System Already Has And this is where I want to land, because this is the part of the argument that matters most. The fear of total AI replacement treats the economy like a one-way machine. AI gets better, humans get replaced, game over. But the economy is not a one-way machine. It is a feedback loop. And feedback loops have brakes. If a company automates too aggressively, it contributes to a decline in the consumer base. As the consumer base shrinks, demand drops. As demand drops, revenue falls. As revenue falls, the stock price falls. As the stock price falls, the trillion-dollar valuation that funded the AI investment in the first place evaporates. At some point, the math forces a correction. Companies need customers more than they need cost savings, because cost savings without revenue is just a slower path to bankruptcy. This is not theoretical. We already see versions of this self-correction in economic history. Companies that cut too deep into their workforce find that they have also cut into their market. Industries that automate too fast create demand gaps that force policy responses, minimum wages, unemployment insurance, retraining programs, and eventually new forms of employment. Full AI replacement is a self-defeating prophecy. It cannot be completed because completing it destroys the conditions that make it valuable. The fear of total replacement ignores this built-in economic brake. It assumes the system will drive off a cliff without anyone noticing that the road has ended. But the road ending IS the signal. Falling demand IS the brake. Disappearing revenue IS the correction. AI is a powerful technology. It will change how we work. It will eliminate some jobs and create others. It will shift the balance between labor and capital in ways that require serious policy attention. All of that is true and all of that deserves real discussion. But the apocalyptic narrative, the one that says AI will replace all humans and take all the money, contradicts the most basic principles of how an economy functions. You cannot run a consumer economy without consumers. You cannot sustain trillion-dollar companies without customers. And you cannot replace the workforce that generates 68% of GDP without collapsing the GDP that makes those companies worth anything in the first place. The question was never whether AI is capable of replacing human labor. The question the fear-mongers refuse to answer is the one Walter Reuther asked seventy years ago: who is left to buy? Sources Federal Reserve (FRED): Personal Consumption Expenditures as % of GDP, Series DPCERE1Q156NBEA [Link] Bureau of Economic Analysis: Consumer Spending Data [Link] Quote Investigator: “How Will You Get Robots to Buy Cars?” Origin and Documentation (2011) [Link] NPR: “The Middle Class Took Off 100 Years Ago…Thanks to Henry Ford?” (2014) [Link] John Maynard Keynes: “Economic Possibilities for our Grandchildren” (1930), via Wikipedia: Technological Unemployment [Link] Conversable Economist: “Automation and Job Loss: Leontief in 1982” (2016) [Link] Foreign Affairs: “Will Humans Go the Way of Horses?” Brynjolfsson & McAfee [Link] Karl Marx: Capital Vol. I, Chapter 25: The General Law of Capitalist Accumulation (1867) [Link] Acemoglu & Restrepo: “Artificial Intelligence, Automation and Work,” NBER Working Paper 24196 [Link] Acemoglu & Johnson: “Rebalancing AI,” IMF Finance & Development (December 2023) [Link] Joseph Stiglitz: “Inequality and Economic Growth,” Columbia Business School [Link] Scientific American: “Unregulated AI Will Worsen Inequality, Warns Nobel-Winning Economist Joseph Stiglitz” (2023) [Link] Fortune: “A huge chunk of U.S. GDP growth is being kept alive by AI spending with no guaranteed return” (Dec 2025) [Link] Fortune: “Thousands of CEOs just admitted AI had no impact” — NBER/PwC Study (Feb 2026) [Link] Fortune: “Without data centers, GDP growth was 0.1%” — Jason Furman, Harvard (Oct 2025) [Link] World Economic Forum: “AI Paradoxes: Why AI’s Future Isn’t Straightforward” (Dec 2025) [Link] Economics Online: The Paradox of Thrift — Keynesian Theory [Link] Corporate Finance Institute: Circular Flow Model — Overview and Economic Implications [Link]

@elonmusk And the back of the truck ia uglier than Rosie O’Donnell 😬

Cybertruck is quicker than a Porsche 911, more powerful than a Ford Raptor

@onlyyou_5510 127

Brain test 98% failed

@edels0n What’s the difference between Iran and Libya?

@ThomasSowell Covid oh boy good times, it showed the humans true nature

The Covid era will be studied as one of the most ridiculous periods in recent history.

Yes but I wanted to make changes to my billing not something I can post on C

@Rogers how is is possible to have that bad of a customer support?

@JonFlynnREstats Looks good to me 👀

This is what you get in Ontario for $1M Guess where?

@funtomvids I would call Justin a lot of names but the good guy would not be one. He single handedly took down our strong Economy with crazy policies and insane woke agenda.

🟥 This photo of our former Prime Minister looking so happy and carefree, gave me comfort that good guys always land on their feet. 👍🏼 All Canadians are happy for you, Justin! 🇨🇦 🙌🏼 We wish you well.

@Timcast What a stupid way to waste money what would a family do with a one time payment of 12k? Versus build and invest in infrastructure projects that generates wealth.

Bernie, lying of course, doesnt explain that net worth is not cash and this would require Bezos and others to liquidate their hard assets which they often are legally able to do and if they did could massively fuck up prices, jobs, and the economy in general

@caitoz Fuck, this much and you can get an STD 😅 just sayin 🤷🏻‍♂️

Fuck the US. Fuck Israel. Fuck Trump. Fuck Netanyahu. Fuck Zionism. Fuck Trump supporters. Fuck the Republican Party. Fuck the Democratic Party. Fuck war. Fuck everyone who helped make this war possible. Fuck the western press. Fuck warmongering think tanks. Fuck the Israel lobby. Fuck the military-industrial complex. Fuck the western intelligence cartel. Fuck the western empire. I hate everyone who inflicted this nightmare upon my species. If you stand by this senseless US-Israeli act of depravity, then I consider you an enemy. And I will never stop reminding everyone of the psychotic agenda you supported. You own this. This is on you. It's on you forever.

@EmilySm43 They deal with women with periods, pregnancy, menopause, mood swings 😉

Women deal with periods, pregnancy, menopause wtf do men deal with ?

@ShaziGoalie That’s scary

$1 million and you can buy a 3,300 sq ft executive home in Oshawa?

@Marlayna29 Amin!

Men, is this true?

@NJBeisner How does someone without ID: 1. Start Business and Registers it. 2. Opens a business bank account 3. Gets a lease of the office or lease of their primary residence. 4. How do they open payment processing accounts. Please provide answers I would Love to know

“I am part of the 40% of Americans that still have not gotten a real ID because I simply have not found the time to wait in line at the DMV for hours on end. And I’m somebody who’s self-employed so imagine if you are a worker who’s hourly and does not have the time or capacity to give a whole day worth of pay to stand in line at the DMV.” This is so outrageous. Unlike this this walking cliché blonde bimbo, I am not blessed enough to be a full-time self-employed content creator, yet somehow I managed to MAKE AN APPOINTMENT and SUBMIT MY DOCUMENTS beforehand online, and I was in and out of the DMV in less than 30. And that’s in Los Angeles, where the DMVs are all hellacious. Every single argument I’ve heard on this is pure fiction and emotionalism. Nobody likes going to the DMV or gathering necessary documents and dealing with bureaucracy, but you can do it. It’s not literally or physically inaccessible to us poors. This argument stems from pure laziness, and it’s a slap in the face to those of us who are actually paycheck to paycheck and are still functioning capable adults.

@realDailyWire Is that Kamala? Didn’t recognize the voice though

@Concern70732755 @grok is this real chart? Is data real?

It just reveals how insignificant our economy is on a global stage on a comparative basis !

@JShamess The median salary for real estate sales agents was $56,320 in 2024. The best paid 25% made $85,440, while the lowest paid 25% made $38,940. I think Military will start looking more and more attractive

There are 10,000 more real estate agents in the Toronto regional board than there are active military troops in Canada. A funny stat that tells a story about Canadian priorities😅