The "Ghost economy" is becoming very real
How a “scary bedtime story” about AI, torched billions in just two days
The Capitalist is a reader-supported publication
Reject Corporate Left Wing Journalism
In the quiet of last Sunday a SubStack post began its rapid ascent across X. It amassed tens of millions of views. By Monday’s market open, the repercussions were visceral.
Shares of IBM plunged 13 percent, marking the company’s worst single-day performance since 2000. Credit-card giants Visa, Mastercard, and American Express shed between 4 and 8 percent of their value. DoorDash, Uber, Blackstone, and KKR followed with drops of up to 8 percent or more. Hundreds of billions were erased from their market capitalizations, with payments companies alone reportedly losing around $100 billion.
The spark was not a disappointing earnings report or ominous Federal Reserve signals. It was a speculative thought experiment from Citrini Research entitled “The 2028 Global Intelligence Crisis.”
Framed as a historical memo from June 30th, in 2028, the fictional report painted a picture of an economy where artificial intelligence had succeeded beyond all expectations—only to unravel the consumer-driven system that had powered the ascent of humankind. “AI didn’t disappoint,” the memo opened provocatively. “It exceeded every expectation. What happened?”
At its core, Citrini’s scenario describes a vicious negative feedback loop. Advanced AI agents, capable of autonomous action, begin displacing white-collar workers en masse starting in late 2025. These high-earning professionals, who disproportionately drive discretionary spending in the economy, see their incomes evaporate or compress as they are forced to compete in gig economies. Consumer spending collapses.
Corporations, flush with higher margins from expanding AI at the cost of workers jobs, pour even more money into AI development, accelerating the displacement. The result is “Ghost GDP”—economic output registered in corporate ledgers and productivity statistics but failing to circulate as human demand for goods, services, restaurants, travel, and housing, because machines do not buy lattes or lease apartments.
In this speculative piece unemployment climbs to 10.2 percent, the S&P 500 falls 38 percent from its 2026 highs, private credit markets seize amid software company defaults, and even prime mortgages in tech hubs like San Francisco begin to crack as incomes falter.
What made this narrative so potent was not its inevitability but its unflinching focus on the demand side of the AI equation. Citrini and co-author Alap Shah explicitly called it a scenario exercise, an exploration of “left tail risks” rather than a forecast, but that nuance didn’t matter to the 27 million people who read it.
For months, markets have been fixated on the supply side of AI: explosive capital expenditures on chips, data centers, bigger models, more compute, and relentlessly higher valuations. NVIDIA’s valuation soared on the promise of AI infrastructure. Hyperscalers announced trillion-dollar data center build-outs. The prevailing narrative was one of boundless productivity and corporate profitability.
Labor-market consequences were often mentioned only in passing, usually as “creative destruction” or “upskilling opportunities.” Citrini flipped the lens.
Citrini dared to ask: Who will buy the output if the buyers are unemployed or underemployed? It asked whether the bullish case on AI capability could itself be bearish for the economy. “What if our AI bullishness continues to be right,” the preface wondered, “and what if that’s actually bearish?” The market, apparently, had not priced that conditional. On Monday, it did, all at once.
The list of industries affected is all encompassing.
Intermediation businesses, those companies that act to smooth the friction generated by humans and businesses interacting, are in the crosshairs first.
Payment platforms, think credit card companies, will be bypassed by AI agentic commerce doing business through stablecoins in seconds.
Delivery platforms, like Uber Eats and DoorDash, will be fragmented by AI agents haggling directly with restaurant’s AI agents over the price of your combination fried rice and spring rolls. A delivery that will show up on your door in an autonomous self driving car driven by, yes you guessed it, an AI agent.
Software as a Service subscriptions, think Zoom, Monday.com and any one of a million “business productivity suites” will die because who needs collaboration software when there are barely any employees to collaborate with.
All these companies and many more not mentioned face existential compression as “zero-friction” AI eliminates human habits that have been the bedrock of the white collar world for a quarter century. The “pain points” of business will be fixed in moments by removing the weakest and most inef`icient link in the chain, the human. This may be great for productivity, but what do we do with all the humans?
This wasn’t mere Luddite panic. It resonated because it echoed emerging data points. According to a 2025 Pew Research Center survey, 52 percent of U.S. workers are worried about AI’s future impact on their jobs, with 32 percent believing it will lead to fewer opportunities. McKinsey Global Institute research projects that activities accounting for up to 30 percent of work hours in the U.S. economy could be automated by 2030, accelerating occupational shifts. The World Economic Forum estimates that AI and automation could displace 92 million jobs globally by 2030 while creating 170 million new ones, for a net gain of 78 million—but with 39 percent of core skills requiring change.
The market’s violent reaction—described by some as an “AI scare trade”—underscored a deeper truth: Wall Street has been salivating over AI capex numbers for months while largely ignoring the potential macroeconomic feedback from labor-market transformation.
Legacy tech like IBM, reliant on services and mainframe modernization, symbolized vulnerability. Delivery and payments firms, built on human-centric models, saw the moats that have protected their profits for nearly five decades questioned. As one widely shared X thread noted, “just five paragraphs on AI agents transacting via stablecoins and bypassing interchange fees appeared to wipe out roughly $100 billion across credit-card stocks alone.”
The backlash was swift, by Wednesday the pushback arrived, as it always does when a narrative threatens consensus. Critics dismissed it as a “scary bedtime story” with shaky economics. They pointed to historical precedents —from the steam engines to computers—where new technology leads to productivity surges and eventually fuels higher living standards and new demand pathways.
Citadel Securities, one of the world’s largest market makers, published a point-by-point rebuttal titled “The 2026 Global Intelligence Crisis.” citing real-time labor data:
AI adoption rates remain stable with no non-linear inflection,
Software-engineering job postings are up 11 percent year-over-year,
New business formation is expanding rapidly,
Construction hiring is buoyed by data-center build outs .
They emphasized that technology diffusion follows S-curves. That organizational inertia and regulation act as brakes, and the elasticity of human desires has always absorbed efficiency gains—echoing Keynes’s 1930 prediction of a 15-hour workweek that underestimated how expanded wants would sustain consumption.
The rebuttal was thorough, data-rich, and institutionally weighty. Yet its very existence underscored the anxiety it sought to dispel. When a major liquidity provider feels compelled to counter a fictional “scary bedtime story” on a SubStack, it signals that the underlying anxieties are probably not fringe. A scenario dismissed as “idiotic” by some traders still moved billions because enough participants feared it might contain a kernel of truth, suggesting that the boundary between fiction and financial reality has already blurred.
Investor psychology is fragile; a narrative that garnered more than 27 million views and that “felt true” moved real capital on Monday and wiped out billions. The scare is being minimized as unfounded, yet the fact that powerhouse firms feel compelled to prove that fact, only heightens the sense that the underlying dynamics could, under certain conditions, prove uncomfortably close to the mark.
Real-world evidence, meanwhile, sits in uneasy tension with the optimists. A January 2026 Harvard Business Review survey of 1,006 global executives found that 60 percent had already made headcount reductions to some degree explicitly in anticipation of AI’s impact—and that was before the technology had delivered commensurate productivity gains.
Stanford’s Digital Economy Lab documented a 13 percent employment decline since late 2022 among workers aged 22–25 in AI-exposed roles—the “canaries in the coal mine.” A Reuters/Ipsos poll from August 2025 showed 71 percent of Americans worried that AI would permanently displace too many workers.
Even OpenAI’s Sam Altman acknowledged in February 2026 that some companies were engaging in “AI washing,” blaming layoffs on the technology when other factors were at play, while also conceding that genuine displacement was coming.
This episode invites a nuanced reckoning. AI’s trajectory may not lead to the precise 2028 cataclysm Citrini laid out, but the speed and scope of which “intelligence abundance” is going to become real, challenges traditional models of economic adaptation. Put plainly, businesses are well known for not adapting to situations as quickly as they should and the AI future is moving faster than anything we have ever dealt with before.
White-collar jobs, once thought insulated, represent over half of employment and the bulk of high-value consumption. Unlike past automation waves targeting routine manual labor, AI targets cognitive tasks, potentially compressing or erasing the “intelligence premium” that underpins the middle class.
The broader implications extend beyond markets to culture and policy.
Work provides not just income but identity and purpose; a “Ghost Economy” risks hollowing out societal fabric. Democracy could strain under inequality exacerbated by AI. Policymakers must grapple with ideas floated in such scenarios: AI compute taxes, expanded direct transfers, massive re-skilling initiatives, or experiments in universal basic services.
Education systems need reinvention to foster uniquely human skills—creativity, empathy, complex problem-solving—that complement rather than compete with machines in a fight they are destined to lose. Businesses face choices: deploy AI for augmentation and value creation or pure cost-cutting that destroys their own customer base.
The Citrini report, in its provocative way, performs a public service by widening the aperture of debate. It reminds us that technology is not destiny; it is shaped by choices. As we stand on the cusp of potentially transformative AI capabilities, the real intelligence test is whether we can foresee and mitigate the dislocations while harnessing the abundance.
History suggests resilience and adaptation, but the velocity of which “the future” is traveling to meet us means that passive optimism may prove insufficient. The ghost in the machine economy is not inevitable obsolescence, but the specter of failing to align technological progress with human prosperity.
How we respond—to the hype, the hazards, and the opportunities—will define not just market cycles, but the contours of 21st-century society.
It certainly was a remarkable reaction to the piece. My personal rebuttal to the underlying idea is that people would much rather deal with other people than with machines in almost every transaction in life. while Gen Alpha may not necessarily feel that way, my take is the rest of the population certainly does.
with that in mind, and understanding how slow corporations are at making transformative moves (after all, otherwise there would never be Clayton Christenson's disruptive innovation with small firms growing while large ones lag changes), AI seems likely to help significantly in some roles (research) and modestly in others (complex legal services) and not at all in others (craftsmanship and physical labor. things will change but not collapse
Seems that faux journalism at CNN is coming to no an end — Ellison just bought it and panic is setting in among the faux journalists there !!