In early 2026 a Substack post spooked Wall Street. Not a Fed statement. Not an earnings miss. A social media post.
The piece was called “The 2028 Global Intelligence Crisis”, written by James van Geelen at Citrini Research. It went viral (16 million views on X), sent IBM shares down 13% in a single session, and prompted famous market bear Michael Burry to quote it with: “And you think I’m bearish.”
With all the attention this received, we thought it would be worth unpacking. Not because we think the market is headed for collapse, but because scary headlines are exactly when our approach matters most: look at the data, look at history, and don’t let a well-written story influence an investment decision.
What Did Citrini Actually Say?
The Citrini piece isn’t a traditional research report. It’s a thought experiment written as a fictional memo from June 2028. You can read the full piece here. The basic chain of events it describes:
- AI rapidly improves at white-collar work: coding, financial analysis, legal research, middle management
- Companies start replacing employees to boost margins
- Displaced workers earn less, spend less, and the consumer economy softens
- The market initially surges as profits rise, but the economy beneath hollows out
- "Ghost GDP" Citrini’s term for output that shows up in national accounts but never circulates through real households
- In their fictional scenario, by June 2028: unemployment hits 10.2%, the S&P has fallen 38% from its highs
To be fair to Citrini, they explicitly state that this a scenario and not a prediction. The risk they raise (productivity gains not flowing to workers) is a legitimate question which deserves discussion. And the writing is compelling enough that markets reacted as if it were a forecast.
What the Data Actually Shows Right Now
Citadel Securities published a direct rebuttal titled “The 2026 Global Intelligence Crisis”, authored by macro strategist Frank Flight. You can read it here. His focus: looking at what is actually happening.
- Unemployment today: 4.28%
- Software engineer job postings: up 11% year over year (Indeed data)
- Daily generative AI usage at work: “unexpectedly stable” according to the St. Louis Fed (little sign of imminent mass displacement)
- New business formation in the U.S.: expanding, partly driven by AI data center construction creating a localized boom in jobs
The Citadel argument isn’t that AI won’t change things, but that the speed and mechanism Citrini assumes simply don’t match the evidence on the ground. Beyond the data, Citadel raises a structural constraint worth understanding; replacing white-collar work at scale isn’t just a software problem. it’s an energy and infrastructure problem.
- Displacing white-collar labour at Citrini’s assumed pace would require orders of magnitude more compute than currently exists
- If automation expands rapidly, demand for chips and energy rises with it, pushing compute costs up
- At some point, the marginal cost of compute exceeds the marginal cost of human labour, and substitution stops
- This creates a natural brake on the kind of runaway feedback loop Citrini describes
Technological diffusion has always followed an S-curve: slow at first, then faster, then plateauing as real-world integration friction reasserts itself. Citrini assumes the exponential phase never slows. History says it always does.
The Optimist’s Version of the Same Story
An excellent read after Citrini’s piece can also be found on Substack. We highly recommend checking out Michael Bloch’s “The 2028 Global Intelligence Boom” published the same day as Citrini’s report. Read it here.
Bloch runs the exact same thought experiment. Same fictional format. Same AI acceleration premise. Opposite conclusion. In his version of June 2028:
- Unemployment: 3.1%
- S&P 500: 12,000
- Real median household purchasing power: up 18% (not from higher wages, but from lower costs)
- Legal fees, financial advice, accounting, and software subscriptions are all dramatically cheaper
- Displaced workers start new businesses faster than in any prior cycle, because AI also collapsed the cost of entrepreneurship
His core argument is simple: when prices fall because production got cheaper, that’s not a crisis, that’s a living standard boom. The intelligence premium didn’t disappear, it deflated. And consumers are the ones who benefit.
This Time Is Not Different, History Tells Us So
This is where our own investment philosophy comes in. We’ve said it before and we’ll keep saying it: we use history as a guide, not headlines.
The fear underneath the Citrini scenario, that a technology might get so good that human work stops mattering, is not new. People asked the same question about:
- The steam engine
- The assembly line
- The personal computer
- The internet
Each time, serious economists concluded the disruption would be permanent and that workers couldn’t adapt fast enough. Each time, they were wrong.
This is because human economic wants have proven effectively unlimited. When productivity rises and costs fall, people don’t stop buying, they buy more and different things. In 1930, John Maynard Keynes predicted that by the early 21st century, rising productivity would mean we’d all be working 15-hour weeks. He was right about productivity growth. He was wrong about everything else, because he couldn’t anticipate how much more people would want once they could afford it.
The record here is unambiguous:
- The automobile did not eliminate jobs, it created entire new industries
- Electrification, computing, and the internet followed the same pattern
- Every major productivity wave has made living standards better, not worse
Betting against this 200-year trend has been the wrong trade, every single time
The Bottom Line
Both the Citrini and Bloch scenarios are theoretically possible. Though it is most likely that neither will fully materialized. What we do know: the historical standard strongly favours an optimistic outcome. Certain data we would pay particular attention to over the coming months and quarters would be:
- New business formation (Census data, 2026): If businesses are being created faster than jobs are being lost, the positive feedback loop is already running
- Services inflation (PCE deflator): If it turns negative, AI deflation is arriving as purchasing power gains
- White-collar unemployment duration: If displaced workers are finding replacement income within 6 months, the consumption spiral Citrini fears has no fuel
Right now, all three of these indicators point in the constructive direction.
Our approach has always been the same: don’t make investment decisions based on thought experiments, no matter how well-written. Prepare portfolios for a range of outcomes. Let data and history inform your positioning, and don’t mistake a compelling story for a forecast.
If you have questions about how we’re thinking about AI, technology, and portfolio positioning, we’re always happy to walk through our thinking.