A response to Citrini Research’s “The 2028 Global Intelligence Crisis” — introducing the actor the original scenario forgot.
Preface
In February 2026, Citrini Research published a remarkable scenario analysis: a fictitious memo from June 2028 describing how AI-driven productivity gains triggered mass unemployment, consumption collapse, and a cascading financial crisis. The S&P 500 fell 38% from its peak. Unemployment hit 10.2%. Prime mortgages — held by borrowers with 780 FICO scores and 20% down payments — started defaulting. The piece was rigorous, vivid, and frightening.
It was also incomplete.
Citrini described the mechanism of the crisis with precision: AI replaces workers, workers stop spending, consumption collapses, financial assets crater. But the scenario treats this process as if it were a natural disaster — an earthquake no one saw coming and no one could steer.
This article introduces the variable Citrini left out: the oligarchies that are not merely surviving the transition, but actively shaping it.
I. The System Was Never Accidental
The conventional narrative frames capitalism as an emergent system — the aggregate result of millions of individual decisions guided by market incentives. This framing is not wrong, but it is dangerously incomplete.
Consider what is empirically documented, not speculated:
- The top 1% of Americans hold 31-32% of total household wealth, up from 23% in the mid-1980s. The top 0.1% — roughly 130,000 families — hold 15%, double what they held in 1978.
- The five richest men in the world doubled their wealth between 2020 and 2024, from $405 billion to $869 billion, while five billion people got poorer.
- Labor’s share of GDP in the United States has fallen from 64% in the 1970s to roughly 56% today. In a $28 trillion economy, that represents $1.7-2.2 trillion per year transferred from labor to capital compared to the 1970s baseline.
- Since 1979, productivity has risen ~70% while median worker compensation has risen only 15-20%. The S&P 500, meanwhile, is up over 4,000%.
- Corporations now return 90-95% of profits to shareholders through buybacks and dividends, up from ~50% in the early 1980s. S&P 500 companies spent $7-8 trillion on buybacks in the decade 2014-2024 alone.
These are not market failures. They are market outcomes — shaped by decades of policy choices on taxation, labor law, financial deregulation, and corporate governance. Choices made not by abstract forces, but by actors with resources to influence the process.
AI is about to accelerate this dynamic to a degree that has no historical precedent.
II. The Most Capital-Intensive, Labor-Light Technology in History
Citrini’s scenario grasps the economic mechanics: when AI replaces a $180,000 product manager with a $200/month agent subscription, the productivity gain flows to the compute owner, not the worker. What the scenario doesn’t explore is what this means for the structure of value creation.
Consider the numbers:
+-----------+-------------------+-------------+------------------+
| Company | Market Value | Employees | Value/Employee |
+-----------+-------------------+-------------+------------------+
| Nvidia | ~$3 trillion | ~32,000 | ~$94 million |
| OpenAI | ~$150-300 billion | ~3,500 | ~$57 million |
| Anthropic | ~$60-100 billion | ~2,000 | ~$40 million |
| Walmart | ~$650 billion | ~2,100,000 | ~$310,000 |
+-----------+-------------------+-------------+------------------+Nvidia generates roughly 300 times more market value per employee than Walmart. AI companies are the most extreme example in economic history of value creation being captured by capital rather than labor. A few thousand employees — and more importantly, a few dozen equity holders — sit atop hundreds of billions in value.
This is not a bug. It is the defining feature of intelligence-as-infrastructure: whoever owns the compute owns the output. And unlike previous technological revolutions, the capital required to compete is so enormous that it creates natural oligopolies.
III. Follow the Money — The 1,000:1 Ratio
If you want to understand what the oligarchies actually believe about the future, ignore their speeches. Read their capital allocation.
What they say (Scenario A: Save Consumer Capitalism)
- Sam Altman (2021): proposed an “American Equity Fund” taxing 2.5% of corporate market value, distributing ~$13,500/year to every citizen. Funded a $60 million UBI study.
- Elon Musk (2024): predicted “universal high income” — not basic, high — enabled by AI and robotics abundance.
- Mark Zuckerberg (2017): “We should explore ideas like universal basic income.”
- World Economic Forum: “Reskilling Revolution” to train 1 billion people by 2030.
- Corporate “upskilling” programs: Microsoft (10 million trained), Amazon ($1.2 billion pledge), IBM (30 million by 2030).
What they do (Scenario B: Acquire Real Assets Before the Collapse)
- Big Tech combined AI capex in 2025 alone: $300-350 billion — Microsoft $80B, Google $75B, Amazon $100B+, Meta $60-65B. Almost entirely in physical infrastructure: data centers, energy, cooling systems.
- Stargate Project: $500 billion over four years (SoftBank, OpenAI, Oracle, Abu Dhabi’s MGX). First site: Abilene, Texas.
- BlackRock’s GAIIP: $100 billion AI infrastructure fund, partnering with Microsoft, MGX, and Nvidia. BlackRock acquired Global Infrastructure Partners for $12.5 billion, becoming the world’s largest infrastructure investor at ~$170 billion in infrastructure AUM.
- Bill Gates: the largest private farmland owner in America — 270,000 acres across 20 states, managed through Cascade Investment.
- Larry Ellison: owns 98% of the island of Lanai, Hawaii — 141 square miles of land, including solar farms and desalination infrastructure.
- Energy acquisitions: Microsoft signed a 20-year deal to restart Three Mile Island for AI power. Amazon bought a nuclear data center campus for $650 million. Google signed deals with Kairos Power for small modular reactors.
- Peter Thiel, Sam Altman, Reid Hoffman: invested in New Zealand property — widely described as “apocalypse insurance.”
- Alternative asset managers: Blackstone ($330B+ in real estate), Brookfield ($100B in renewables/transition), Apollo, KKR — total alternative AUM has grown from $7-8 trillion in 2013 to $25-27 trillion in 2025.
The ratio: for every dollar flowing toward redistribution or worker transition, roughly one thousand dollars flow toward acquisition of real assets, compute infrastructure, and energy.
The speeches are Scenario A. The capital flows are Scenario B.
IV. The Oligarchic Playbook
If we treat the capital flows as revealed preference — what oligarchies actually believe will happen — a coherent four-part strategy emerges:
1. Own the Intelligence Infrastructure
If human intelligence was the scarce input that gave labor bargaining power, and machine intelligence is replacing it, then owning compute is owning the new labor force. This is not metaphorical — it is the literal economic logic.
The combined AI infrastructure spending announced for 2025-2029 exceeds $1.5 trillion. This is the largest coordinated capital deployment into physical assets since the post-WWII reconstruction, and it is being executed not by governments but by a handful of private actors and sovereign wealth funds.
2. Convert Financial Wealth to Real Assets Before the Crisis
Citrini’s scenario describes the S&P 500 falling 38-57% from peak. Private credit defaults cascade. Mortgage-backed securities come under stress. 401(k) balances evaporate.
But this destruction is not evenly distributed. Those who have already converted financial assets into ownership of land, energy, water, compute infrastructure, and essential services retain value through the crisis. Those still holding equities, pension funds, and 401(k)s — primarily the middle class — absorb the losses.
This is precisely the pattern of 2008-2012: financial crisis destroyed middle-class housing wealth, after which institutional investors (Blackstone, Invitation Homes) bought hundreds of thousands of foreclosed homes at distressed prices, becoming the largest landlords in America.
The playbook is identical. The scale is larger.
3. Control the Pace of Policy Response
The most valuable asset during the transition is time. Every month that redistribution is delayed, the capital conversion continues and labor’s bargaining position weakens further.
Consider the lobbying asymmetry: in 2023, the tech sector spent over $70 million on lobbying in the US alone. Meta: $20M. Amazon: $21M. Google: $14M. Microsoft: $11M. OpenAI, which didn’t exist as a lobbying entity two years prior, spent $1.5M and growing rapidly.
During the EU AI Act negotiations (2022-2024), tech companies met with EU officials hundreds of times to soften provisions. The resulting Act regulates risk categories and transparency — it does not address wealth redistribution from AI at all.
Meanwhile, in the US, Biden’s October 2023 AI Executive Order — which only studied labor impacts, without creating any displacement fund — was rescinded by the Trump administration in January 2025.
The pattern: accelerate deployment, decelerate regulation. Not because anyone is coordinating a conspiracy, but because every actor is independently incentivized to do exactly this.
4. Frame the Narrative
The dominant narrative around AI follows a specific script:
- “AI will create new jobs” — the same promise made in every industrial revolution, applied without evidence that it holds when the technology substitutes general intelligence rather than specific manual tasks.
- “Workers need to upskill” — shifting responsibility from systemic design to individual adaptation. Amazon’s $1.2B “upskilling” program trains workers for Amazon’s own needs — it is labor pipeline maintenance, not redistribution.
- “It’s too early to regulate” — while $300B+ flows into infrastructure annually.
- Public debate focuses on AI bias, deepfakes, and existential risk — important issues, but none of them address the central economic question: who owns the output of machine intelligence?
The question that is never asked at Davos is the only one that matters.
V. The Geopolitics of Oligarchic AI
Citrini’s scenario is implicitly American. But the oligarchic transition is global, with different power structures pursuing parallel strategies:
The American Model: Tech Billionaire Oligarchy
A handful of individuals — Altman, Musk, Zuckerberg, Bezos, Nadella, Pichai — control the trajectory of AI development through private companies. Their power derives from capital and talent concentration. The US government acts as enabler (CHIPS Act, Stargate) rather than director.
Strength: speed, innovation, capital abundance. Vulnerability: no democratic mandate, concentrated decision-making, social instability risk.
The Chinese Model: Party-State Oligarchy
The CCP treats AI as an instrument of state power. All major Chinese AI companies (Baidu, Alibaba, ByteDance, DeepSeek) operate under Party oversight. The social credit system deploys AI for population management through 700 million surveillance cameras.
DeepSeek’s January 2025 demonstration — frontier-competitive models trained at a fraction of US costs — showed that algorithmic efficiency can partially offset hardware restrictions, undermining the US export-control strategy.
Strength: coordination, long-term planning, no internal opposition. Vulnerability: innovation constraints, demographic decline, chip supply dependency.
The Gulf Model: Sovereign Wealth Conversion
Saudi Arabia and the UAE are executing the most explicitly strategic play: converting finite hydrocarbon wealth into permanent compute and AI infrastructure.
- Saudi PIF ($930B): invested in Stargate, launched “Alat” with $100B mandate for electronics/semiconductors, building NEOM as an AI-first city.
- Abu Dhabi’s MGX: anchor investor in both the BlackRock GAIIP and Stargate. G42 received $1.5B from Microsoft.
- Logic: oil revenues fund the transition before fossil fuel demand declines. They are running a clock.
Strength: capital abundance, no democratic accountability, clear strategic vision. Vulnerability: dependency on Western technology partners, small population, geopolitical exposure.
The European Non-Model
The EU has no oligarchic AI champion. Mistral AI (~$6B valuation) is Europe’s best hope, dwarfed by US and Chinese competitors. EU AI venture capital ($8-10B/year) is roughly one-eighth of US levels ($60-70B+).
The EU regulates through the AI Act. But regulation without competitive capability may produce the worst outcome: rules that European companies must follow while American and Chinese oligarchies capture the value.
The uncomfortable insight: every region with a powerful AI sector has an oligarchic structure driving it. The EU’s more democratic, distributed approach is producing the weakest AI capability. Power concentration and AI dominance appear to be structurally linked.
VI. The Paradox Citrini Identified — And Its Resolution
The original article identifies a devastating feedback loop: AI replaces workers → workers stop spending → consumption collapses (70% of GDP) → even AI-driven companies lose revenue → crisis.
This raises the question: doesn’t the oligarchy need consumers?
The answer depends on what you consider wealth.
If wealth is financial assets — yes, they need consumers.
Stock prices, bond yields, and real estate valuations all depend on economic activity that ultimately traces back to consumer spending. In this world, the oligarchy must accept some form of redistribution — a UBI, tax credits, consumption vouchers — to maintain demand.
But note the terms: the redistribution would be controlled. Enough to buy, never enough to accumulate. Consumers are maintained as dependents, not as potential capitalists. A digital feudalism with Prime delivery.
If wealth is real assets — no, they don’t.
If you own the land, the energy, the water, the compute, and the AI that runs everything — you don’t need consumers. You need resources. The economy bifurcates: an automated circuit serving those who own the infrastructure, and a subsistence economy for everyone else.
In this world, the financial crisis Citrini describes is not a catastrophe for the oligarchy — it is the transition mechanism. Financial assets crash, destroying middle-class wealth (pensions, 401(k)s, home equity). Those who already converted to real assets emerge on the other side owning a larger share of everything that matters.
The capital flows suggest the oligarchies are preparing for this second scenario while publicly advocating for the first.
VII. The Transfer Is Already Underway
The 2028 crisis, if it occurs, will not be the beginning of the wealth transfer. It will be its culmination. The transfer is already happening:
Phase 1 — Now (2024-2026): Convert and Consolidate
- $300-350B/year flowing into physical AI infrastructure
- Alternative asset AUM tripling in a decade to $25-27 trillion
- Farmland, water rights, energy assets being acquired at scale
- Lobbying to delay regulation and redistribution
Phase 2 — Crisis (2027-2028): Financial Destruction, Real Asset Preservation
- Financial markets crash (Citrini’s scenario: S&P -38 to -57%)
- Middle-class wealth evaporates (401(k), home equity, pension funds)
- Real asset values (land, energy, compute) remain structurally supported by physical utility
- Distressed assets become available for acquisition at deep discounts
Phase 3 — Post-Crisis (2029+): New Rules
- Whoever controls the real assets defines the post-crisis economic architecture
- Some form of redistribution emerges (UBI, consumption credits) — on terms set by asset owners
- The consumer capitalism of 1945-2025 is replaced by something that doesn’t yet have a name
VIII. The Canary and the Mine Owner
Citrini’s article ends with an evocative line: “The canary is still alive.”
The implication is that there is still time to build proactive frameworks — redistribution mechanisms, policy responses, social safety nets — before the feedback loops achieve escape velocity.
This is true. But it misses a crucial dimension.
The canary is still alive. But the question is not whether we can hear it. The question is: who owns the mine?
And the data suggests that the mine owners are not waiting for the canary to die. They are buying the next mine.
The productivity gains of artificial intelligence could be the greatest shared prosperity event in human history — or the greatest concentration of wealth and power. The technology does not choose. The oligarchies already have.
Data Sources and Methodology
This analysis draws on publicly available data from: Federal Reserve Distributional Financial Accounts, World Inequality Database, Bureau of Economic Analysis (NIPA), Bureau of Labor Statistics, Economic Policy Institute, Oxfam inequality reports, SEC corporate filings (DEF 14A proxies), S&P Global Market Intelligence, Preqin Global Alternatives Reports, Bain & Company Luxury Market Studies, OpenSecrets lobbying data, EU Transparency Register, company earnings reports and investor presentations, WEF Future of Jobs Reports, and sovereign wealth fund annual reports.
The original scenario analysis referenced is “The 2028 Global Intelligence Crisis” by Citrini Research and Alap Shah, published February 22, 2026.