On February 22, 2026, a finance writer named James Van Geelen published a thought experiment on Substack. By Monday morning, the Dow had fallen 800 points. Enterprise software stocks were in freefall — CrowdStrike down 10%, Zscaler down 9.9%, Intuit down 8%. Gold surged past $5,200 an ounce. All because of a fictional memo written from the future.
Here’s what you need to know.
What This Memo Actually Is
Citrini Research’s “The 2028 Global Intelligence Crisis” is not a prediction. It’s a stress test. Written as if it were a financial report from June 2028, it asks one uncomfortable question: What if everything we’re excited about with AI... actually goes right?
And what if that’s the problem? ( most people are not thinking about it that way )
The memo imagines a world where AI delivers on every promise - better code, smarter agents, faster everything - and then traces what happens to real people, real companies, and real markets when human intelligence is no longer the scarce resource the entire economy was built around.
Think of it like a fire drill. Nobody is saying the building is currently on fire. But the drill is designed to show you if people know where the exits are.
The Simple Version of What It Describes
Here’s the chain of events the memo lays out, step by step:
AI gets really good at white-collar work. By late 2025, agentic coding tools let a decent developer replicate a mid-market SaaS product in weeks. Companies start asking: “Why are we paying $500K/year for this software when we can build it ourselves?”
Companies cut workers and buy AI instead. This is rational for each individual company. Cut headcount, save money, invest in AI tools, maintain output with fewer people. Margins expand. Earnings beat. Stocks rally.
But nobody asks who’s buying the products. White-collar workers make up 50% of U.S. employment and drive roughly 75% of discretionary consumer spending. When they lose their jobs or take massive pay cuts, they stop spending. Machines, it turns out, spend zero dollars on dinner, vacations, or mortgage payments.
A feedback loop begins. Workers get laid off → they spend less → companies lose revenue → companies invest more in AI to protect margins → more workers get laid off. Citrini calls this the “Intelligence Displacement Spiral” - a negative feedback loop with no natural brake.
The financial system starts to crack. The memo traces how this cycle eventually threatens the $13 trillion U.S. mortgage market (because prime borrowers suddenly can’t earn what they used to), the $2.5 trillion private credit market (because PE-backed software companies start defaulting), and even credit card networks (because AI agents route payments around interchange fees using stablecoins).
In the memo’s fictional 2028, the S&P 500 has fallen 38% from its October 2026 highs. Unemployment sits at 10.2%. The economy is in recession.
Why It Calls This “Ghost GDP”
This is one of the memo’s most powerful concepts, and it’s worth understanding.
Imagine a single GPU cluster in North Dakota producing the same output as 10,000 office workers in Manhattan. On paper, GDP looks great — productivity is booming, output is surging, the national accounts are glowing.
But none of that productivity circulates through the real economy. Those 10,000 workers used to spend their salaries on rent, groceries, childcare, restaurants, and car payments. The GPU cluster spends on electricity.
That’s Ghost GDP: “Output that shows up in the national accounts but never circulates through the real economy”. The headline numbers look fine. The economy underneath is hollowing out.
Why This Matters More Than a Typical Doom Piece
The standard response to AI job fears is: “Technology always destroys jobs and creates new ones. ATMs didn’t kill bank tellers. The internet created more jobs than it destroyed.”
The memo directly addresses this — and explains why it might not apply this time.
Every prior technology disrupted specific tasks but still needed humans to do the new work. AI is different because it’s a general intelligence that improves at the very tasks displaced humans would redeploy to. A coder displaced by AI can’t simply move into “AI management” because AI is increasingly capable of that too.
There’s a second reason: speed. Previous technology transitions played out over decades. The internet took 15 years to fully reshape the job market. AI capabilities are improving every quarter, and each improvement accelerates the displacement.
Nobel laureate Daron Acemoglu put it bluntly: “If we go down this path of destroying jobs and creating more inequality, U.S. democracy is not going to survive”. MIT economists Acemoglu and David Autor both argue that whether AI becomes a crisis or a transition depends almost entirely on speed — not on whether the gains eventually arrive.
What the Experts Are Saying
The memo landed into an already charged environment. Here’s where key voices stand:
The Worried Camp
IMF Managing Director Kristalina Georgieva said at Davos 2026 that AI is “like a tsunami hitting the labor market,” especially in advanced economies where 60% of jobs are at risk. She warned that “most nations and businesses are not equipped for this”.
Anthropic CEO Dario Amodei has publicly predicted that AI will eliminate 50% of entry-level white-collar jobs within one to five years, with unemployment potentially reaching 10-20%.
Geoffrey Hinton, the “Godfather of AI” and Nobel Prize winner, warned that 2026 specifically will bring a new wave of AI-driven job losses.
Federal Reserve Governor Michael Barr outlined three scenarios for AI’s labor market impact, including a “jobless boom” where “AI agents replace or displace a range of professional and service occupations” and “a large share of the population is essentially unemployable.” He warned that “society would have to rethink the social safety net”.
Goldman Sachs estimates 300 million jobs globally will be affected by AI by 2028. McKinsey estimates up to 30% of hours currently worked could be automated by 2030.
The Skeptical Camp
Oxford Economics published a report in January 2026 arguing that “firms don’t appear to be replacing workers with AI on a significant scale.” The firm suggested companies may be using AI as a cover for routine headcount reductions: “We suspect some firms are trying to dress up layoffs as a good news story”.
ServiceNow CEO Bill McDermott fired back directly: “The speculation of AI will eat software companies is out there. Let’s clear it up with the facts.” He argued that enterprise AI depends on workflow orchestration platforms like his rather than replacing them.
Deutsche Bank prompted a proprietary AI tool to forecast displacement, and it predicted 92 million jobs eliminated by 2030 — but 170 million new roles created.
Michael Bloch wrote a formal companion piece to Citrini’s memo — same premise, same fictional format, opposite conclusion. His “2028 Global Intelligence Boom” argues that deflation returns savings to consumers, displaced workers start businesses (7.2 million new applications in his scenario), and purchasing power rises even as nominal wages flatten. His core counterargument: the bears confuse the repricing of one sector with the collapse of the entire economy.
The Nuanced Middle
Fed Governor Barr himself acknowledged that the most likely scenario is gradual adoption, where “many workers successfully retrain and retain their jobs or find new ones.” But he stressed he could not rule out the extreme scenario and that policymakers should prepare for it.
The Atlantic summarized the debate this way: Both optimists and pessimists agree that what matters is not whether the gains from AI eventually arrive — it’s whether they arrive fast enough to prevent a destabilizing gap in the middle.
What Most People Will Miss
Here are the insights buried in the memo that most people will skim past — and yet they matter the most.
1. This feedback loop doesn’t slow down when the economy weakens
In a normal recession, the cause eventually self-corrects. Overbuilding slows, rates fall, construction restarts. But AI investment isn’t cyclical capex that gets cut during downturns. It’s OpEx substitution. A company spending $100M on employees and $5M on AI shifts to $70M on employees and $20M on AI. AI spending increases even as total spending shrinks. The engine of disruption accelerates precisely when the economy needs it to slow down.
2. “Permanent capital” is actually your neighbor’s retirement savings
The memo reveals how private equity firms like Apollo, KKR, and Brookfield bought life insurance companies and turned annuity deposits — regular people’s retirement savings — into fuel for private credit deals. When PE-backed software companies default, the losses don’t land on sophisticated Wall Street investors. They land on the annuity policies of Main Street households. The “permanent capital” that was supposed to make the system resilient was regular Americans’ nest eggs.
3. Rate cuts cannot fix a structural problem
In 2008, the Fed could cut rates and buy mortgage-backed securities because the crisis was about financial conditions. This crisis, the memo argues, is about the real economy engine — AI making human intelligence less scarce and less valuable. Cutting rates to zero won’t change the fact that a Claude agent can do the work of a $180,000 product manager for $200 a month. Traditional monetary policy tools address the symptoms, not the disease.
4. Incumbents accelerated their own destruction
The historical model says incumbents resist new technology and lose to nimble startups (Kodak, Blockbuster, BlackBerry). That’s not what the memo describes. Instead, threatened companies became AI’s most aggressive adopters — because they couldn’t afford not to. ServiceNow cut headcount and used the savings to fund the very technology disrupting it. Each company’s response was individually rational. The collective result was catastrophic.
5. India’s entire economic model is under threat — and almost nobody is talking about it
Buried in the memo is a devastating aside: India’s IT services sector — $200 billion annually, the backbone of its current account surplus — is existentially threatened because the marginal cost of an AI coding agent has collapsed to the cost of electricity. TCS, Infosys, and Wipro face accelerating contract cancellations. The rupee drops 18%. The IMF begins “preliminary discussions” with New Delhi. This has massive geopolitical implications that extend far beyond Wall Street.
6. AI agents are already routing around the financial system’s toll booths
The memo describes AI shopping agents discovering that stablecoins on Solana or Ethereum L2s are cheaper than the 2-3% interchange fees credit cards charge. In machine-to-machine commerce, there’s no loyalty, no habit, no convenience — just optimization. If this plays out, it threatens the revenue model of Visa, Mastercard, American Express, and every card-issuing bank. The market got a taste on Monday: AXP fell 7.7%, MA dropped 3.7%.
The Real Question the Memo Asks
Strip away the fictional framing, the market panic, and the expert debates, and the memo is asking one fundamental question:
Every institution in our economy — the labor market, the mortgage market, the tax code, the financial system — was built for a world where human intelligence was scarce. What happens when intelligence is not scarce?
Human intelligence derived its value from scarcity. When it was rare, people could charge a premium for it. That premium paid mortgages, funded consumer spending, generated tax revenue, and kept the entire system spinning.
The memo argues we’re watching the beginning of that premium unwinding — and that the institutions built around it haven’t caught up yet.
Whether this plays out as Citrini’s crisis or Bloch’s boom depends on variables nobody can fully predict: the speed of AI capability improvement, the speed of human adaptation, the effectiveness of policy response, and whether deflationary benefits reach consumers fast enough to offset income losses.
But the question itself? That one is already real.
Read the Full memo here…