99 files. 18 resources. 18 countries. 20 cascade models. 5 probability-weighted scenarios.

Chat with this research | Browse the site

A bottom-up geopolitical simulation of the 2026 Iran War (US/Israel vs Iran, started Feb 28), built entirely from open-source intelligence. Unlike top-down political punditry, this analysis starts from physical resource constraints — oil flows, mineral supply chains, semiconductor inputs, food systems — and models how disruptions cascade across industries, countries, and each other simultaneously.

Built using Claude as a network of parallel research agents, with each resource, country, and cascade analyzed independently, then integrated into a unified simulation.

Most war analysis focuses on military operations and diplomacy. This project asks different questions:

• What happens when 33% of global helium goes offline and chip fabs can't run?
• What happens when 67% of the world's bromine (Dead Sea, Israel/Jordan) is in a war zone and every circuit board needs flame retardants?
• What happens when fertilizer shipments through Hormuz stop and the Northern Hemisphere has already planted with insufficient inputs?
• What happens when all of these hit simultaneously?

The answer: cascades compound in ways that no single-issue analysis captures. That's what this project models.

China is the war's strategic winner. It controls rare earth processing, is insulated from Hormuz disruption, and holds a November 2026 deadline on gallium/germanium exports that gives it maximum leverage over the US mid-war. Every week the war continues, China's relative position strengthens.

The chip famine is structural, not transient. Three independent semiconductor inputs — helium (33% offline), bromine (67% at risk), neon (combinatorial risk) — are disrupted simultaneously. Combined with Taiwan's 11-day LNG reserve, chip production faces a 30-50% output cut that no single-source workaround can fix.

The food crisis is already locked in. Northern Hemisphere spring planting (March-May 2026) is happening with disrupted fertilizer supply. Even if the war ends tomorrow, Q3-Q4 crop yields will disappoint. 1 billion+ people face food insecurity by late 2026.

November 2026 is the convergence point. US midterm elections (Nov 3), China's gallium/germanium export ban deadline (Nov 27), and Europe's winter energy crisis (Dec 1) all collide. This is when maximum political, economic, and strategic pressure peaks simultaneously.

Insurance outlasts the war. Lloyd's war-risk designations persist weeks to months after ceasefire. The $1:$390 trade multiplier means insurance disruption freezes far more economic activity than the physical damage alone.

The analysis is organized in seven intelligence layers, each building on the one below:

Layer 7: COMMUNICATION        Public assets, outreach, distribution to external audiences
Layer 6: BLIND SPOTS          What we don't know and might be wrong about
Layer 5: SIMULATION            Phase-by-phase predictions and scenarios
Layer 4: CASCADES              How disruptions compound across domains  ← the analytical core
Layer 3: INDUSTRIES            Sector-specific impact (chips, food, pharma, defense, auto, energy)
Layer 2: ACTORS                Country-by-country incentive analysis
Layer 1: RESOURCES             Production, prices, trade flows, stockpiles

Every claim cites the layer below it. Simulation cites cascades. Cascades cite industries and resources. Resources cite primary sources with dates.

• Parallel research agents deployed per resource, per country, and per cascade — not a single narrative thread
• All claims sourced from major news outlets, think tanks, and government reports (Feb-March 2026)
• Analysis built bottom-up from resource constraints, not top-down from political assumptions
• Verified facts distinguished from reasonable inferences distinguished from speculation
• Combinatorial analysis identifying correlated disruption cascades that single-issue analysis misses

For journalists, researchers, and content creators, the following public-facing assets are available:

All assets are licensed under CC BY 4.0. Feel free to adapt and reuse with attribution.

This entire repository was built using Claude Code (Anthropic's CLI agent) as an orchestration layer. The process:

1. Parallel deep dives: 8+ research agents deployed simultaneously, each investigating a single resource, country, or cascade in depth
2. Bottom-up integration: Findings aggregated into cascade models that identify compounding interactions
3. Simulation synthesis: All layers combined into probability-weighted scenarios
4. Continuous updates: As the war evolved day-by-day, new intelligence was integrated and cascades re-traced

The project's knowledge management uses an "Agentic Brain" architecture — a structured system of intelligence layers, navigational indexes, and dependency tracking that allows AI agents to efficiently navigate and update 76 interconnected files.

• All analysis is based on open-source reporting as of March 24, 2026
• Wars are inherently unpredictable — a single escalation could redirect everything
• This is an analytical framework, not a crystal ball
• Resource data has reporting lags; some figures may be outdated by days or weeks

This work is licensed under CC BY 4.0. You're free to share and adapt it with attribution.