Citrini’s “2028 Global Intelligence Crisis” spooked markets and drew a White House response. Strip out the doom and it reads as a map of which business models break when AI agents kill switching costs.
Citrini Research’s viral memo imagines a 2028 where AI displaces white-collar work fast enough to crash the S&P 500 by 38% and push unemployment above 10%. It is a scenario, not a forecast. The durable signal underneath is narrower and far more useful for anyone running a company: agentic AI collapses switching costs, and friction-moat business models are the ones exposed.
Here is what almost everyone who amplified this memo got wrong. It was never a prediction. Citrini wrote it as a piece of financial fiction, a “macro memo from June 2028,” and the internet did what the internet does. It treated a thought experiment like a weather report, panicked, and then argued about the wrong thing. That misreading is exactly why the useful part got buried.
What Happened
In late February 2026, Citrini Research published a scenario titled “The 2028 Global Intelligence Crisis.” Written from the imagined vantage point of June 2028, it describes a self-reinforcing “intelligence displacement spiral”: AI capabilities improve, companies need fewer workers, white-collar layoffs climb, displaced workers spend less, margin pressure pushes firms to invest more in AI, and the loop tightens again.
Then it went off. The piece racked up roughly 16 million views after Michael Burry amplified it, drew a public response from the White House, and knocked real money around. IBM fell about 13% and weakness spread across software, payments, and delivery names. Citadel Securities published a formal rebuttal calling the whole thing a misread of macro fundamentals.
The timing matters for why this is back in the feed. Goldman Sachs has now put a real number on the actual labor effect, and it is nothing like the memo. Economist Pierfrancesco Mei expects US unemployment to drift from 4.3% to about 4.5% by the end of 2026, with faster AI adoption adding at most another 0.3 points. His phrasing for the observed damage so far: visible but moderate. Meanwhile the Challenger layoff report shows AI became the single most-cited reason for job cuts in March 2026. The scare is real. The catastrophe is not showing up in the data.
The Backstory
Citrini Research is a macro analysis shop founded in 2023 by James van Geelen. This memo, co-authored with Alap Shah, was deliberately built as a narrative device, not a base case. Its most quotable idea is “Ghost GDP,” a world where output shows up in corporate profit but does not circulate, because the displaced workers who used to spend that income no longer have it. Since consumer spending drives roughly 70% of GDP, the memo argues, you get profit without prosperity and a deflationary spiral that ordinary monetary tools cannot fix.
It is elegant, internally consistent, and built to travel. That is the problem. A clean single-variable story spreads faster than a messy accurate one, and this one was engineered for exactly that.
How the Spiral Is Supposed to Work
Cut through the macro theater and the memo rests on one genuinely sharp micro-observation. AI agents reduce switching costs. When an agent can complete a task, compare providers, or replicate a workflow on demand, the friction that used to lock customers in starts to evaporate. That friction was never a neutral detail. For a huge share of software and services businesses, it was the moat.
Once switching costs fall, intermediation reprices. The middleman who profited from being embedded in a process, rather than from being the best at it, loses pricing power. That is the mechanism that actually deserves attention, and it has nothing to do with whether unemployment hits 10% in 2028.
The Business Model Angle
This is where the memo stops being macro fiction and becomes a business-model stress test. Read it as a hit list and it sorts companies into two piles.
The exposed pile is friction-moat businesses: SaaS products with weak differentiation whose pricing power depends on being wedged into a workflow an agent can route around rather than through. If a customer stayed because leaving was annoying, not because the product was irreplaceable, that revenue is now on borrowed time. The same logic hits workflow-embedded intermediaries and “accessible” middlemen whose value proposition was convenience and lock-in.
The resilient pile is businesses agents have to run through, not around: the systems of record, the deeply embedded workflows, the products with real differentiation or data that competitors cannot cheaply reproduce. Owning the subscription rails is not enough. Owning a position an agent cannot substitute is.
The tell will show up in the metrics before it shows up in the stock. Net revenue retention is the number to watch, because switching-cost erosion bleeds out through expansion and churn long before it hits the headline. Founders who track the right upstream SaaS signals will see the repricing coming. The ones staring only at ARR will get surprised. Note that even a giant like IBM, the stock the market punished hardest on the memo, sits across both piles at once depending on which segment you look at.
The Risk
Now the case against taking any of this literally, because it is strong. Citadel Securities argues the memo’s core error is conflating recursive technology with recursive economic adoption. The model may improve every quarter. Companies do not restructure every quarter. Org inertia, retraining lag, regulation, and plain human friction all sit between a capability and a pink slip.
The real world keeps supplying evidence for the skeptics. Klarna cut hundreds of roles for an AI chatbot, then admitted it went too far and started rehiring for the complex work the AI could not handle. The Yale Budget Lab found no macro-level signal of mass displacement yet, with the rate of occupational change basically flat. Even Citrini hedges its own thesis with a line worth remembering: repricing is not the same as collapse. Economies find new equilibria. Painful is not the same as terminal.
The honest read is that the mechanism is plausible and the timeline and magnitude are the fiction. Treat the memo as a scenario to stress-test your model against, not a countdown clock.
Quick Questions
Is the Citrini report a real prediction? No. It is a constructed scenario written from an imagined June 2028, meant to explore one possible pathway, not to forecast a specific outcome.
Did it actually move markets? Yes. After Michael Burry amplified it, the piece drew roughly 16 million views, a White House response, and a visible selloff in software and payments stocks, with IBM down about 13%.
What does the current data actually show? Modest effects. Goldman Sachs projects US unemployment rising only from 4.3% to about 4.5% through 2026, and calls AI job losses visible but moderate so far.
What is the real business lesson? Switching costs are repricing. Business models whose moat was friction and lock-in are exposed. Models an AI agent has to run through, rather than around, are far more defensible.
The Business Model Analyst Take
The 2028 crash memo is a bad forecast and a very good stress test. As doom, it fails: the actual 2026 labor data is closer to a rounding error than a spiral, and its author never claimed otherwise. As a business-model diagnostic, it is one of the sharpest things published this cycle, because it points a floodlight at the single most overrated moat of the SaaS era, which is customer inertia.
The strategic move is not to brace for 2028. It is to audit your own model against one question the memo forces you to ask: when an AI agent can do the switching, the comparing, and the busywork your customers used to tolerate, what is left that they would still pay you for? If the honest answer is “it would be a hassle to leave,” you do not have a moat. You have a countdown. The companies that use this scare to find that answer early will be the ones the spiral never touches.
Reporting based on Citrini Research’s “2028 Global Intelligence Crisis,” with market and labor data from Goldman Sachs, the Challenger layoff report, the Yale Budget Lab, Citadel Securities, and B Capital.
