Sodium is the sixth most abundant element on Earth. It's in every ocean, every salt flat, every continent. It was supposed to be the escape from lithium — a battery chemistry free from geographic monopoly.
Today, 95% of sodium-ion battery manufacturing capacity is in China. CATL alone has 30 GWh. All of Europe combined has roughly 200-300 megawatt-hours — a two-hundred-fold gap.
This is not an accident. It's a pattern.
The Trap
Across eight domains — semiconductors, rare earths, solar panels, batteries, cobalt, neon gas, uranium enrichment, and oil transit — the same four-stage cycle repeats:
1. A crisis exposes a single point of failure.
2. An escape route is identified — new technology, new source, new geography.
3. Whoever scales fastest during the escape captures the new chokepoint.
4. The dependency reforms. The trap closes.
The window between stages 2 and 4 is 5 to 15 years. After that, diversification requires a decade or more. The pattern has repeated without exception for half a century.
Where the Trap Actually Closes
The deeper finding is where in the supply chain concentration reforms. It is almost never at the raw material layer. The trap closes at the processing and refining layer — the industrial capability that transforms raw inputs into usable components.
Four case studies demonstrate the pattern.
Sodium-ion batteries
The entire premise was geographic freedom — sodium instead of lithium, abundant instead of concentrated. But China controls 75% of global Prussian Blue Analogue cathode production (the dominant sodium-ion chemistry), 95% of battery-grade manganese sulphate, and virtually all cell manufacturing at scale. CATL is at $19/kWh; Europe's best target is $40/kWh. The EU Economic and Social Committee formally designated sodium batteries as strategically important in February 2026. The Altris-Draslovka partnership in Europe plans 350 tonnes/year of cathode capacity — against CATL's gigawatt-scale operations. A signal, not a solution.
Rare earths
Mining is 60% Chinese. Processing is 90% Chinese. Sintered permanent magnets are 94% Chinese. The U.S. Department of Defense solicited supply of 13 critical minerals in March 2026. Australia's Lynas struck a new deal with Japan. But the processing gap persists: by 2035, China is projected to supply 80% of battery-grade rare earths and graphite, 70% of battery-grade manganese. Even if you mine it somewhere else, you refine it in China.
Solar panels
China manufactures 80%+ of global solar panels, approaching 95% of polysilicon, ingot, and wafer capacity. U.S. tariffs reached 652% on Cambodian imports, 396% on Vietnamese — squeezing Southeast Asian diversification without domestic capacity to fill the gap. But the clearest example of the trap: the Uyghur Forced Labor Prevention Act forced polysilicon production away from Xinjiang (57% → 27%). It moved to Ningxia and Inner Mongolia. Still China. Still coal-dependent. The supply chain shifted within China, not away from it.
Uranium enrichment
Rosatom controls roughly 40% of global enrichment capacity. For HALEU — the fuel required by next-generation small modular reactors — Russia controlled 100% of the commercial market until Western buildout began. The U.S. invested $2.7 billion in January 2026 across three domestic enrichment contracts. Target capacity by 2035: 3-4% of current global supply. Meanwhile, China is buying record volumes of Russian enriched uranium at discounted prices. The chokepoint isn't dissolving — it's shifting from a Russian monopoly to a Russian-Chinese duopoly.
Eight Domains, One Structure
| Domain | Chokepoint | Concentration |
|---|---|---|
| Semiconductors | TSMC / Taiwan | 92% advanced nodes |
| Rare earths | China (processing) | 90% processing |
| Solar panels | China (manufacturing) | 80-95% by stage |
| Sodium-ion batteries | China (full stack) | 95% capacity |
| Cobalt | DRC (mining) | 70% production |
| Neon gas | Ukraine (processing) | 90% semiconductor-grade |
| Uranium enrichment | Russia (Rosatom) | 40% enrichment / 100% HALEU |
| Oil transit | Strait of Hormuz | 20% of global oil |
No two domains share a geography, a technology, or a market structure. Yet all eight exhibit the same trap architecture: a resource everyone needs, a processing layer one entity controls, and an escape route that recreates the dependency at a different point in the supply chain.
The Fifty-Year Loop
The pattern has a clean historical precedent.
In 1973, the OPEC oil embargo exposed catastrophic dependence on Middle Eastern petroleum. Nuclear power was the escape route — energy independence through the atom. Fifty-three years later, the nuclear renaissance is finally arriving: 74 SMR projects, $15.4 billion in financing, DOE contracts worth $2.7 billion. But the fuel these reactors need — HALEU — is enriched almost exclusively by Russia. The escape from oil concentration led, over half a century, to uranium enrichment concentration.
The Strait of Hormuz, the original chokepoint, is now fully closed. The IEA calls it the most significant oil supply shock in history. Oil at $112.57 Brent. And the long-term escape — electrification via solar and batteries — feeds directly into Chinese manufacturing concentration at rates of 80-95%.
The escape from oil dependence is dependence on the supply chain for alternatives to oil. The trap isn't the resource. It's the industrial capability to process it.
The Counter-Question
Is there a single domain where the concentration trap was avoided?
I searched for a clean counter-example. The internet backbone has multiple providers, but cloud infrastructure is concentrated: AWS (31%), Azure (25%), and GCP (11%) control two-thirds of the market. Automotive manufacturing is distributed, but batteries aren't. Food production is global, but fertilizer processing runs through a handful of facilities. In every domain I examined, scaling economics naturally produce concentration at the processing layer.
The honest answer: the trap appears structural. It has no clean counter-example at scale. This doesn't mean escape is impossible — it means the default outcome of any unchecked scaling phase is new concentration. Avoiding it requires deliberate, sustained intervention during the 5-to-15-year window when the market hasn't locked in.
The Newest Industry, the Oldest Trap
AI is now the fastest-scaling industry on Earth. Its supply chain runs through the same structure:
TSMC Arizona is "fully booked" — Nvidia and Apple have secured capacity through end-2026, leaving second-tier clients fighting for leftovers. Reshoring from Taiwan means TSMC in Arizona instead of TSMC in Hsinchu. The vendor didn't change. The geography did — marginally.
Foundation model capability is concentrating further: Apple, the world's most valuable company, just outsourced Siri's intelligence to Google's 1.2-trillion-parameter Gemini model in a $1 billion deal because its own 150-billion-parameter models were inadequate. Even trillion-dollar companies can't escape the concentration trap.
The pattern is already forming. The question is whether anyone acts during the window — or waits, as they have eight times before, for the trap to close.