Some thoughts after reading Citrini Research’s piece “The 2028 Global Intelligence Crisis.”
Citrini Research recently published a piece that went viral across the investing world. Framed as a “macro memo from June 2028,” it imagines a doomsday scenario where large-scale AI displacement of white-collar labor triggers an economic collapse — the S&P 500 down 38% from its peak, unemployment spiking to 10.2%, consumer spending falling off a cliff, and a wave of mortgage defaults. The piece was influential enough that it’s been pointed to as one of the catalysts behind the broad tech-stock selloff on February 24 — Michael Burry reposted it on X with the comment “Still think I’m just bearish?”, and that day IBM dropped nearly 13%, with DoorDash, American Express, and KKR all down more than 8%.
I read the whole thing carefully. It’s genuinely well written, with a tight chain of logic: AI raises productivity → companies cut headcount → incomes fall → spending contracts → companies lean on AI even harder to cut costs further → a self-reinforcing “intelligence substitution spiral” → eventually a systemic financial crisis. The piece calls this “phantom GDP” — output that shows up in the national accounts but no longer flows through the real economy.
Clever as the argument is, I don’t share its conclusion.
1. Technological revolutions have never shrunk the economy — they’ve always expanded it#
Citrini’s core assumption is that once productive capacity grows far faster than income and demand can keep up, the economic structure loses its stability. That sounds reasonable, but history keeps proving the opposite.
The Industrial Revolution replaced manual labor — by the same logic, textile workers, blacksmiths, and coachmen all lost their jobs, so consumption should have collapsed. What actually happened: collapsing production costs created entirely new consumer markets, and new factories, new jobs, and new industries kept appearing one after another, and humanity’s material abundance took a genuine leap forward.
The computing and internet revolution replaced typists, mail carriers, telephone operators, bank tellers — but it gave rise to the entire internet industry, e-commerce, social media, and the mobile app economy. Behind every wave of “replacement” was a wave of “creation” ten or a hundred times its size.
The key economic logic: technological progress isn’t a zero-sum game — it expands the total size of the economy. Higher productivity makes goods and services cheaper, freeing up purchasing power that flows into new categories of consumption, creating demand that simply didn’t exist before. Over the past 200 years, every time someone has predicted “machines will put humanity out of work and collapse the economy,” the actual result has been an economy that grew multiple times over, with total employment rising rather than falling.
Citrini’s argument makes a classic mistake: mistaking the friction of a transition period for the final outcome. AI adoption will genuinely cause structural unemployment in parts of the economy in the short term, and white-collar work will be hit first and hardest. But that’s a transitional phase, not a permanent state. When AI pushes the cost of writing code, building spreadsheets, or drafting legal documents down toward zero, what actually happens isn’t “nobody makes money anymore” — it’s “far more people can now do far more things they couldn’t do before.”
That’s also why I’m currently holding a heavy position in AI stocks — not because I’m ignoring the risk, but because I believe we’re in the early stage of a technological shift on the same scale as the Industrial Revolution or the internet. Panic narratives come and go, but the underlying, fundamental boost technology gives to productivity doesn’t reverse.
2. What AI brings isn’t equality — it’s an unprecedentedly fast reshuffling of class#
If I’m skeptical of Citrini’s “economic crisis” thesis, I’m genuinely worried about the distribution problem AI is creating.
I used to believe in AI as an equalizer. AI would give everyone access to a top-tier coding assistant, translator, writing partner, legal advisor — technology, education, and cognitive ability all becoming more equal across the board.
I now think that was naive. Equalization only works if everyone’s using the same AI. In reality, some people are using Claude Opus 4.6 to write production-grade code, while others are stuck with a free model producing something half-finished — and the output gap between the two starts at 10x, easily.
My own experience is the clearest example. When I first started using AI tools, I’d hesitate over a $3 secondhand Cursor subscription. Now, I pay for rising monthly AI subscriptions without a second thought — not because I’ve gotten that much richer, but because I’ve genuinely internalized that the price of a token is the price of productivity.
The logic behind this is fairly harsh: someone with access to frontier models can produce a production-ready solution in one pass and create value a thousand times over; someone without the means to pay is stuck with free or cheap models, spends three hours producing something half-baked, and only ever creates one unit of value. What someone else finishes in minutes, you might spend 100 hours on and still not match in quality — and the practical problem is that a weaker model tends to make code messier with every edit rather than better. That’s not a difference in effort — it’s a difference in tooling generations, and it’s a rout.
The reality in 2026: model capability is diverging faster, and token prices are climbing faster. The gap between the best models and the worst isn’t linear — it’s exponential. Which means the productivity gap between people who can afford top-tier AI and people who can’t is going to widen at a pace we haven’t seen before.
3. The world ahead: paradise for the wealthy, a much harder climb for everyone else#
Zooming out: from industrialization to the information age to the intelligence age, society has gone through three major waves of labor displacement. Industrialization replaced physical labor — assembly lines replaced artisans. The information age replaced repetitive mental labor — ERP systems replaced bookkeepers, search engines replaced librarians. The intelligence age is now replacing what used to be considered untouchable — high-level mental labor: AI can write code, make diagnoses, produce investment strategies, even write research papers.
Each wave redefined who could actually earn a living. This one is especially disruptive, because it’s aimed squarely at what used to be the most stable, highest-earning group in society: white-collar knowledge workers.
Citrini’s concern isn’t baseless — AI genuinely will disrupt the white-collar job market in the short term. But the crisis narrative misses the other side of it: the people who master AI will have their ability to create value amplified enormously.
Tomorrow’s winners won’t be the people who hold some specific piece of expertise — because AI is commoditizing knowledge itself — but the people who know how to direct AI. An ordinary person who’s good with AI can out-produce ten experts who aren’t. That’s not an exaggeration; it’s already happening. But the flip side is just as real: people without AI skills will lose more than just an efficiency edge — they’ll lose access to more and more jobs outright. Once a company realizes one AI-literate employee can do the work of five, the other four positions simply stop existing.
That points toward a polarized world: people who can master AI and afford to pay for it will see their quality of life amplified like never before — better healthcare, better education, better investment returns, more creative leverage. People without AI skills, or without the means to pay for AI, will find it harder and harder just to get by — not only will job opportunities keep shrinking, but the barrier to catching up will keep rising too.
Closing thought: the real risk isn’t that AI is too powerful — it’s that you haven’t kept up#
Citrini’s piece paints a doomsday picture of AI collapsing the economy. I think the more realistic picture is different: the economy won’t collapse, but how wealth gets distributed is going to be completely rewritten.
AI won’t make everyone poorer, but it will make some people extraordinarily wealthy while leaving others far behind. This isn’t an economic crisis — it’s a reshuffling of class, and it’s happening faster than any before it.
As someone living through this shift firsthand, my advice is simple: embrace AI, get in early, and don’t wait. Looking back, you may have missed the wave of economic reform, missed Bitcoin when nobody wanted it early on, missed the early growth of the internet — not acting now means missing AI’s own founding year.
