107 lines
7.5 KiB
Markdown
107 lines
7.5 KiB
Markdown
---
|
|
type: archive
|
|
source: "Citrini Research (Alap Shah / James Val Geelen)"
|
|
url: https://www.citriniresearch.com/p/2028gic
|
|
date: 2026-02-22
|
|
tags: [rio, ai-macro, labor-displacement, private-credit, financial-crisis, scenario-analysis]
|
|
linked_set: ai-intelligence-crisis-divergence-feb2026
|
|
status: processed
|
|
claims_extracted:
|
|
- "AI labor displacement operates as a self-funding feedback loop because companies substitute AI for labor as OpEx not CapEx meaning falling aggregate demand does not slow AI adoption"
|
|
- "Incomplete digitization insulates economies from AI displacement contagion because without standardized software systems AI has limited targets for automation and no private credit channel to transmit losses"
|
|
- "Private credits permanent capital is structurally exposed to AI disruption through insurance-company funding vehicles that channel policyholder savings into PE-backed software debt"
|
|
- "White-collar displacement has lagged but deeper consumption impact than blue-collar because top-decile earners drive disproportionate consumer spending and their savings buffers mask the damage for quarters"
|
|
---
|
|
|
|
# THE 2028 GLOBAL INTELLIGENCE CRISIS — Citrini Research
|
|
|
|
Speculative macro memo written from the perspective of June 2028, describing a bear scenario for AI's economic impact. Published Feb 22, 2026. Went viral and moved markets — triggered a risk-off move wiping billions in market cap on Feb 23. Citadel Securities published a rebuttal.
|
|
|
|
## Core Thesis
|
|
|
|
AI displaces white-collar workers through an OpEx substitution feedback loop with no natural brake. Unlike cyclical recessions that self-correct, AI capability improvement is self-funding: companies lay off workers, save money, buy more AI, lay off more workers. The engine that caused the disruption got better every quarter.
|
|
|
|
## Key Mechanisms
|
|
|
|
### Ghost GDP
|
|
"The output was growing. The income wasn't." Productivity surging while gains flow to capital and compute, not labor. GDP growing while the real economy deteriorates because the circular flow of income — households earn, spend, firms earn, hire — breaks when firms replace hiring with AI subscriptions.
|
|
|
|
### The Intelligence Displacement Spiral
|
|
- White-collar workers displaced first (product managers, consultants, customer service, software)
|
|
- Displaced workers downshift to service/gig economy, compressing wages there too
|
|
- "Sector-specific disruption metastasized into economy-wide wage compression"
|
|
- Still-employed professionals spend defensively, reducing consumption further
|
|
- Autonomous delivery/driving then displaces the gig workers who absorbed the first wave
|
|
|
|
### OpEx Substitution (No Natural Brake)
|
|
- AI investment is OpEx substitution, not CapEx addition
|
|
- Company spending $100M employees + $5M AI becomes $70M employees + $20M AI
|
|
- AI budget 4x'd while total spend fell 15%
|
|
- Falling aggregate demand does NOT slow AI buildout — it's self-funding
|
|
- "The intuitive expectation was that falling aggregate demand would slow the AI buildout. It didn't."
|
|
|
|
### Top-Decile Consumption Concentration
|
|
- Top 10% of earners account for 50%+ of all consumer spending
|
|
- Top 20% account for ~65%
|
|
- White-collar displacement hits the demand base for the entire discretionary economy
|
|
- 2% decline in white-collar employment = 3-4% hit to discretionary consumer spending
|
|
- Lagged impact: savings buffers mask damage for 2-3 quarters, then consumption craters
|
|
|
|
### Private Credit Contagion
|
|
- Private credit grew from <$1T (2015) to >$2.5T (2026)
|
|
- Heavily deployed into software/tech deals at valuations assuming mid-teens revenue growth in perpetuity
|
|
- PE-backed software LBOs entered restructuring when ARR stopped recurring
|
|
- Moody's downgraded $18B of PE-backed software debt across 14 issuers (Apr 2027)
|
|
- Zendesk: $5B direct lending facility marked to 58 cents — largest private credit software default on record
|
|
|
|
### The Insurance Channel
|
|
- "Permanent capital" backing private credit was actually life insurance policyholder money
|
|
- Apollo/Athene, KKR/Global Atlantic, Brookfield/American Equity — alt managers acquired life insurers as funding vehicles
|
|
- Annuity deposits invested into PE-originated private credit
|
|
- Fee-on-fee perpetual motion machine that worked under one condition: the private credit had to be money good
|
|
- When software loans defaulted, losses hit balance sheets holding policyholder savings
|
|
- Offshore SPV structures (Bermuda/Cayman reinsurers) created opacity — "who actually bore the loss was genuinely unanswerable in real time"
|
|
- "A daisy chain of correlated bets on white collar productivity growth" — Fed Chair Warsh
|
|
|
|
### Mortgage Impairment
|
|
- $13T residential mortgage market built on assumption borrowers remain employed at roughly current income level
|
|
- Not subprime: 780 FICO, 20% down, verified income — "bedrock of credit quality"
|
|
- "In 2008, the loans were bad on day one. In 2028, the loans were good on day one. The world just...changed."
|
|
- Delinquencies spike in SF, Seattle, Manhattan, Austin — tech/finance heavy ZIP codes
|
|
- National average stays within historical norms, but trajectory is the threat
|
|
|
|
### Policy Impotence
|
|
- Traditional toolkit (rate cuts, QE) addresses financial engine but not real economy engine
|
|
- "The real economy engine is not driven by tight financial conditions. It's driven by AI making human intelligence less scarce and less valuable."
|
|
- Federal receipts running 12% below CBO baseline — payroll and income tax receipts falling
|
|
- Labor's share of GDP dropped from 56% (2024) to 46% — "sharpest decline on record"
|
|
- "The government needs to transfer more money to households at precisely the moment it is collecting less money from them in taxes"
|
|
- Proposed "Transition Economy Act" and "Shared AI Prosperity Act" (sovereign wealth fund / compute tax) stuck in partisan gridlock
|
|
|
|
### The Intelligence Premium Unwind
|
|
- "For the entirety of modern economic history, human intelligence has been the scarce input"
|
|
- Every institution — labor market, mortgage market, tax code — was designed for a world where intelligence was scarce
|
|
- Machine intelligence is now a competent substitute across a growing range of tasks
|
|
- "Repricing is not the same as collapse" — but nobody's framework fits because "none were designed for a world where the scarce input became abundant"
|
|
- "Whether we build them in time is the only question that matters"
|
|
|
|
## Key Data Points (fictional, scenario-based)
|
|
- S&P 500: -38% peak-to-trough
|
|
- Unemployment: 10.2%
|
|
- Initial jobless claims: 487,000 (highest since April 2020)
|
|
- India IT services: rupee fell 18% in four months as services surplus evaporated
|
|
- Labor share of GDP: 56% → 46%
|
|
- Federal receipts: 12% below CBO baseline
|
|
|
|
## Disclosure
|
|
- Written Feb 2026 as scenario analysis, not prediction
|
|
- "We are certain some of these scenarios won't materialize"
|
|
- "The premium on human intelligence will narrow" — this they consider certain
|
|
- Co-authored with Alap Shah of LOTUS
|
|
|
|
## Connections to Knowledge Base
|
|
- Validates mechanism in [[LLMs shift investment management from economies of scale to economies of edge]] — same force destroying incumbent intermediaries
|
|
- Directly relevant to Belief #5 (legacy intermediation is rent-extraction) — but shows disruption can be net negative on 3-5 year horizon
|
|
- OpEx substitution mechanism challenges [[internet finance generates 50 to 100 basis points of additional annual GDP growth]] — the GDP may grow but not route through households
|
|
- "Technology exponential, coordination linear" gap — Citrini argues it's become unbridgeable on relevant timescale
|
|
- Private credit channel connects to [[optimization for efficiency without regard for resilience creates systemic fragility]]
|