1. *"The US has aircraft carriers. China has the periodic table."* The US won the military war in hours. But China controls rare earth processing, gallium/germanium for defense electronics, cobalt refining, lithium refining — and is insulated from Hormuz. By year-end, China establishes its first permanent Gulf naval presence and yuan-denominated oil rises 30-40%. The war's winner never fires a shot.
2. *Late April pharma cliff.* Pharmaceutical buffer stocks exhaust ~60 days after Hormuz closure. By late April, antibiotic and metformin shortages begin globally. China controls 70% of India's API imports. WHO's Dubai logistics hub is paralyzed. This one hits in 4 weeks.
3. *The 24-day November gauntlet.* Between Nov 3 (US midterms, 53% oppose the war) and Nov 27 (China's gallium/germanium export ban suspension expires), China holds maximum leverage over a politically wounded US — while Europe faces winter with gas storage at 55-70% (target: 90%). Three independent pressure peaks in 24 days.
4. *Taiwan's 11-day clock.* Taiwan has 11 days of LNG reserves. China doesn't need to invade — just delay tankers as "enhanced customs inspections." Combined with helium shortage hitting TSMC fabs, the chip crisis isn't caused by military strikes. It's caused by thermodynamics.
5. *1 nuclear state becomes 5 in a decade.* Even the best-case scenario leaves Iran with 440.9 kg of 60% enriched uranium as a permanent threshold state. This alone triggers Saudi acquisition via Pakistan (6-24 months). Then Turkey (3-7 years), Japan (6-12 months if it decides), South Korea (1-2 years). The NPT doesn't survive this war.
I built this using Claude Code as an orchestration layer — parallel research agents deployed per resource and per country, each producing a self-contained deep dive with sourced data. Cascade models then identify how disruptions compound across domains simultaneously.
The motivation: most war analysis is top-down political commentary. But wars are resource disruption events. If you model from physical constraints upward — who produces what, what ships through where, what breaks when supply X goes offline — you get different predictions than pundit-level analysis.
Findings that surprised me:
1. The chip famine is three independent input crises hitting simultaneously (helium 33% offline + bromine 67% at risk + neon combinatorial) — no single-source fix exists
2. The food crisis was locked in by March. Northern Hemisphere spring planting happened with disrupted fertilizer supply. Even an immediate ceasefire can't fix Q3-Q4 yields.
3. November 2026 is a convergence point where US midterms, China's gallium/germanium export deadline, and Europe's winter energy crisis all collide.
4. Insurance is a bigger weapon than missiles. The $1:$390 trade multiplier means Lloyd's war-risk designations freeze far more economic activity than physical damage.
All sourced from major news outlets, think tanks, and government reports (Feb-March 2026). Built with Claude Code using an "Agentic Brain" knowledge architecture — the CLAUDE.md in the repo documents how 76 files are organized and interconnected.