The Spice Must Flow: Why Taiwan is Our Dune
The future of AI and how it might break the economic theory of peace
The first missile hit Lahore at 3:47 AM on May 4th. By breakfast, the Dell Theory of Conflict Prevention was dead.
For fifty years, we've bet civilization on a simple idea: countries that trade together don't fight together. We shipped millions of jobs overseas as a down payment on permanent peace. We called it globalization. We said it was inevitable.
Then India and Pakistan—nuclear powers, democracies, economic partners—spent four days proving we were fools.
Not border skirmishes. Not sanctions. Actual missiles into actual cities. Cyber attacks on power grids. Drones swarming across borders. Two countries with everything to lose choosing economic destruction over mutual dependence.
Taiwan watched every minute. So did China. So did the engineers at TSMC who've been briefed on "Protocol 7"—the plan to reduce the world's most advanced semiconductor fabs to molten silicon rather than let them fall into Chinese hands.
If the Dell Theory broke over a territorial dispute, would it hold for the most valuable substance in the world?
The Spice
There's a science fiction novel that explains our current predicament better than any Pentagon briefing. In Frank Herbert's Dune, the entire galaxy depends on a single substance—melange—that can only be produced on one desert planet. Without the spice, interstellar travel stops. Civilization collapses.
We're living in that novel. The spice is semiconductors. The planet is Taiwan.
But here's what Herbert got wrong: in Dune, everyone understands the spice's value. They fight wars over it. They scheme and murder for it.
In our world, we pretended the spice would keep the peace.
The Numbers That Matter
TSMC produces 92% of all chips below 7nm—the only chips that matter for AI. One company. One island. One target.
Without these chips, you can't build competitive AI systems. Without competitive AI, you lose the technological race. And unlike previous technological races, this one has a peculiar property: whoever wins might win forever.
For forty years, we told ourselves this concentration was actually a good thing. Economic mutually assured destruction. China needs Taiwan's chips, so China won't invade Taiwan. The Dell Theory of Conflict Prevention, wrapped in silicon.
Last month, India and Pakistan lit that theory on fire.
What Kashmir Taught China
On April 22, terrorists killed 26 tourists in Kashmir. Mostly Western backpackers, their final Instagram stories still showing wildflowers and mountain peaks. Within 72 hours, two nuclear powers were calculating whether $10 billion in annual trade was worth more than national security.
Security won.
This isn't a story about who was right or wrong in Kashmir. It's about what their conflict teaches us about Taiwan. Both nations made rational choices: India had every right to respond to terrorism, Pakistan felt compelled to defend its sovereignty. That's exactly the problem.
The escalation ladder was a masterclass in how rational nations make impossible choices:
Day 1: India blames Pakistan-based militants, demands extradition
Day 3: Pakistan refuses, claiming insufficient evidence, mobilizes border forces
Day 7: Indian special forces attempt a cross-border raid targeting terror camps
Day 11: Pakistani F-16s crater the runway at Pathankot Air Base
Day 12: Indian missiles hit Lahore
Not border skirmishes. Not "surgical strikes." Actual conventional warfare between nuclear powers who share rivers, railroads, and until that moment, deeply integrated economies.
The Dell Theory Dies in Lahore
Thomas Friedman promised us that no two countries with McDonald's would ever fight. The Dell Theory went further: countries in the same supply chain can't afford war. The math is simple—war costs more than any possible gain.
Pakistan's textile industry depends on Indian cotton. Indian pharmaceuticals need Pakistani raw materials. Their economies are welded together by a thousand supply chains built over decades.
Four days. That's how long those welds held under pressure.
By the time the ceasefire was signed on May 7th, the damage was:
78 confirmed dead
$10 billion in trade relationships: severed
The Indus Waters Treaty (survived three previous wars): suspended
The Dell Theory of Conflict Prevention: dead
But here's what should terrify you: six weeks later, the trade borders remain closed. The economic damage is permanent. Both countries are poorer, weaker, more isolated.
And neither country regrets it.
The Lesson
In Beijing, war planners did the math. If India and Pakistan would torch $10 billion over Kashmir—worthless mountains that produce nothing—what would America do for Taiwan? What would China do?
Taiwan produces 92% of humanity's advanced chips. Not textiles. Not pharmaceuticals. The foundation of every AI system, every advanced weapon, every technological edge that will define the next century.
The terrifying lesson of Kashmir isn't that nations fought over something important. It's that they destroyed $10 billion in trade over strategically worthless mountains.
Taiwan isn't worthless. Taiwan is everything. If territorial disputes over strategically marginal land can override economic rationality, what chance does the Dell Theory have against the actual future of technological civilization?
The Knowledge Problem
Semiconductors are catastrophically different from every resource war in history: you can't just capture fabs and start making chips.
You can steal chip designs. You can buy the same ASML lithography machines. You can poach engineers. But you still can't make competitive chips. Because the knowledge isn't in any of those things—it's in the accumulation of ten thousand tiny discoveries about what works.
TSMC engineers know that vibrations at 3.2 Hz affect the 7nm process differently than the 10nm process. They know which clean room air patterns minimize defects for each specific chip architecture. They know the exact temperature fluctuations that turn a 99% yield into a 60% yield.
None of this is written down. It can't be. It exists as institutional memory across thousands of engineers who've spent decades turning the impossible into the routine.
This is why Intel, despite inventing the microprocessor, fell behind. It's why China, despite throwing hundreds of billions at the problem, remains stuck at 14nm. It's why even Samsung, with all of South Korea's engineering talent, can't match TSMC's yields.
But here's the interesting part: TSMC's Arizona plant is proving something important. American workers are successfully producing 4nm chips, with 3nm coming online next year. The same processes that seemed impossibly complex are being mastered by engineers in Phoenix.
What changed? TSMC sent their knowledge, not just their machines. Hundreds of their senior engineers are rotating through Arizona, teaching the accumulated wisdom of thirty years. They're transferring the unwritten rules—which vibrations matter, which temperatures are critical, how to read patterns in yield data that aren't in any manual.
This proves the constraint isn't American capability. It's access to knowledge. With proper knowledge transfer, we can build the most advanced chips in the world. We just need TSMC to share their crown jewels—the bleeding-edge 2nm and beyond processes they're keeping in Taiwan. And they won't share those without security guarantees.
The knowledge problem makes semiconductors fundamentally different from oil or any other resource. You can rebuild oil infrastructure. You can't rebuild thirty years of accumulated manufacturing wisdom. But you can transfer it—if you have the right incentives.
The Dune Strategy
In Dune, the Fremen's power doesn't come from controlling the spice. It comes from their ability to destroy it. "The power to destroy a thing is the absolute control over it," Paul Atreides realizes.
Taiwan understands this perfectly. They've made it explicit: if China invades, they'll destroy the fabs. Every advanced semiconductor facility will be rubble before Chinese forces can secure them.
It's brilliant as deterrence goes. China wants those fabs. They need them to compete in AI. Why would they invade if it means destroying the very thing they're trying to capture?
But Taiwan's strategy has the same flaw as the Dell Theory. It assumes rational calculation in an irrational situation.
Consider China's actual position:
They're already locked out of advanced chips by U.S. export controls
Every month, they fall further behind in AI development
At current rates, they'll be permanently relegated to technological second-tier status
Now consider their options:
Accept permanent technological inferiority
Invade Taiwan, destroy the fabs, reset everyone to zero
Hope something changes
After watching India and Pakistan choose option 2 over Kashmir, a territory whose strategic value couldn't justify the economic destruction, which option do you think China will choose over semiconductors?
Why This Time Is Different
The thing about technological races is that most of them have catch-up mechanics. Fall behind in steel production? Build more furnaces. Miss the first wave of the internet? Copy what works and iterate faster.
AI doesn't work that way. And that's because of a mathematical truth that changes everything: intelligence improvements compound exponentially.
Here's a simplified model. Country A has access to cutting-edge 3nm chips. Country B is stuck with 7nm technology. Both are racing to build artificial general intelligence.
Year 1: Country A's AI runs 5x faster, helping them design chips 10% more efficiently. Year 2: Those better chips let them build AI that designs 30% faster. Year 3: That AI creates architectural breakthroughs Country B can't even understand. Year 5: Country A's AI is designing post-silicon quantum systems while Country B struggles with basic optimization.
The gap doesn't close. It explodes.
This is why semiconductors are genuinely different from every previous resource war. Oil gave you temporary advantages. Manufacturing capacity could be rebuilt. Even nuclear weapons could be eventually matched.
But falling behind in the AI race might be permanent. The country that achieves artificial general intelligence first doesn't just get a temporary advantage. They get the ability to ensure no one else ever catches up.
In game theory terms, we've moved from finite games to infinite ones. And infinite prizes justify infinite costs.
The Broken Bargain
Here's what makes this especially bitter for Americans: we already paid the price for peace.
For forty years, we've watched American manufacturing jobs disappear. Factory towns became ghost towns. Middle-class families saw their futures exported to Asia. We were told this was progress—that a "service economy" was more advanced than making things.
But the real bargain was different: we were trading jobs for peace. Economic interdependence would make war impossible. Your job was the price of ensuring your children never faced conscription.
The Dell Theory wasn't just an academic argument. It was the promise made to the American middle class. Give up your economic security, and we'll guarantee your physical security.
India and Pakistan just proved this bargain no longer holds.
Not only did economic interdependence fail to prevent their war, but the economic damage persists after the shooting stopped. Six weeks later, $10 billion in trade remains dead. The jobs lost aren't coming back. The trust is gone.
We hollowed out the Midwest for a promise of peace. Now we're going to get the war anyway. Except this time, we'll be fighting over chips we no longer know how to make.
The Path Forward
The solution is simple but politically impossible: give Taiwan an explicit security guarantee.
Not strategic ambiguity. Not weapons sales. An actual, public commitment that invasion means war with America.
This will cause problems. China will be furious. They'll threaten economic retaliation. They might even increase military provocations.
So what? They're already doing all of those things. We've already paid the economic price through Trump's tariffs. The Dell Theory is already dead. We might as well act like it.
But a security guarantee alone isn't enough. We need three parallel efforts:
1. Immediate Protection Tell Taiwan publicly that we'll defend them. Make it impossible for China to believe they can take the island without fighting America.
2. Knowledge Transfer TSMC has shown they'll share knowledge when protected—they're already producing 4nm chips in Arizona. With a security guarantee, we can accelerate transfer of their most advanced processes. Not just the machines—the knowledge.
3. Complete Embargo No chips above 14nm to China. No manufacturing equipment. Secondary sanctions on any country that helps them circumvent restrictions. Yes, this will accelerate tensions. But we're already in a race where second place might mean permanent subjugation.
The alternative is to wait and hope. Hope that strategic ambiguity holds. Hope that economic interdependence matters more than technological supremacy. Hope that China makes different calculations than India and Pakistan.
That's not a strategy. It's wishful thinking.
The Lesson of May 2025
The India-Pakistan War of May 2025 lasted four days. In historical terms, that's nothing. But it might be remembered as the moment we learned that the post-Cold War order was based on a fantasy.
The fantasy was that economic interdependence made war obsolete. That rational actors would always choose prosperity over conflict. That the arc of history bent toward peace through trade.
India and Pakistan—democracies, nuclear powers, economic partners—proved this wrong. They chose security over supply chains. They picked sovereignty over prosperity. They burned ten billion dollars in trade over territorial rights to mountains most of their citizens couldn't find on a map.
Now imagine the same calculation, but the stakes are exponentially higher. Instead of Kashmir's mountains, it's Taiwan's fabs. Instead of securing borders, it's securing permanent technological dominance. Instead of protecting territory, it's controlling the future of human civilization.
The spice must flow. But unlike Dune, we can't afford to let the spice planet burn. We need those fabs. We need that knowledge. We need to act like we're in the race we're actually in—not the peaceful economic competition we wish we were in.
Every month we delay is another month China watches the gap grow wider. Another month they calculate whether destroying the fabs is better than being permanently locked out. Another month closer to someone making the same miscalculation India and Pakistan just made.
The difference is that when the shooting stops this time, we won't just lose trade relationships. We'll lose the ability to build the technologies that define the next century. And unlike trade, that's not something you can rebuild with a peace treaty.
This was written before the Iran conflict escalated.
The Iran strikes are exactly the type of conflict that traditional theory would predict (adversaries with no economic ties). The whole point of the Dell Theory is that economic interdependence between trading partners should prevent war.