Highlights

• Yttrium prices outside China have surged from single digits to over $100-300/kg due to export controls, creating a bifurcated market with stable pricing inside China and volatile scarcity elsewhere—transforming a theoretical vulnerability into an active market crisis.
• AI infrastructure depends on a complex network of rare earth materials beyond yttrium alone, including neodymium, dysprosium, and terbium across cooling, semiconductors, and power systems—meaning no single bottleneck fix can resolve the systemic risk.
• The critical mismatch: AI infrastructure scales on 2-5 year timelines while rare earth supply chains require 7-15 years to build, and China's dominance in midstream processing—not just mining—creates durable leverage that Western policy has failed to address.

---

The race for artificial intelligence supremacy is not being fought in code. It is being fought in the periodic table.

A recent essay in The National Interest (opens in a new tab) by Morgan Bazilian, Jahara Matisek, and Macdonald Amoah correctly identifies yttrium as a quiet choke point—an obscure rare earth underpinning the turbines powering America’s AI buildout. Their thesis is clean: hyperscale data centers need power; power needs turbines; turbines depend on yttrium-based coatings. China controls the supply. Therefore, China holds leverage. Accurate—but already outdated.

The authors correctly identified yttrium as a strategic chokepoint—but events have moved faster than their framework. What was framed as a future vulnerability has already become an active market dislocation, with prices outside China surging dramatically and supply tightening under policy-driven export controls. Their linear model—AI → power → turbines → yttrium—misses a more complex reality in which yttrium sits within a web of interdependent, non-substitutable materials across cooling, semiconductors, and advanced systems. Crucially, this is no longer just a supply issue but a structural market shift: a bifurcated system where China maintains controlled pricing and access while ex-China markets face volatility and scarcity. Hence, prices can change quickly. With AI infrastructure scaling rapidly and supply chains taking a decade or more to build, the timing gap is now acute. In short, the chokepoint the tri identified is not emerging—it has already been activated, revealing a deeper truth: in today’s rare earth economy, access—not just availability—is the real source of power.

From Chokepoint to Shock

As Rare Earth Exchanges™ reported in “Yttrium Shock,” this is no longer a theoretical vulnerability. It is a live market fracture. Yttrium prices outside China have surged from single-digit dollars per kilogram to well over $100/kg (even $300) in recent transactions—an extraordinary repricing driven not by demand, but by policy. Export controls and tightening flows have created what can only be described as a two-tier market:

• Inside China: stable, state-influenced pricing
• Outside China: volatile, scarce, and increasingly strategic

This is not a commodity cycle. It is a structural split.

The Illusion of a Single Bottleneck

Even now, focusing on yttrium alone misses the broader system risk.

AI infrastructure depends on a full stack of rare earth-enabled subsystems—cooling, sensors, power electronics, and control systems. Yttrium appears not only in turbine coatings but also in ceramics, phosphors, and thermal management architectures embedded throughout data centers.

The implication is stark: securing yttrium for turbines does not solve the problem. Constraints in adjacent materials—neodymium, dysprosium, terbium—can still halt deployment.

This is not a supply chain. It is a dependency network with no single point of failure—and no easy workaround.

China’s Advantage Is Industrial

The deeper issue is not geological. It is industrial.

China’s dominance extends across separation, refining, alloying, and manufacturing of yttrium-based materials such as yttria-stabilized zirconia. That is where leverage becomes durable.

Western policy remains misaligned—fixated on upstream mining while the real bottlenecks sit in midstream and downstream capabilities. Ore is not the sole constraint. Processing is key.

Performance Degradation: The Hidden Risk

The most underappreciated consequence of yttrium scarcity is not immediate shutdown—but degradation.

Jet engines run hotter. Semiconductor yields slip. Turbines wear faster. Precision systems lose reliability.

In defense terms, this translates into lower readiness, longer maintenance cycles, and weakened deterrence—not because systems fail outright, but because they perform worse over time.

The Time Mismatch

AI infrastructure is scaling over a 2–5-year timeline. Rare earth supply chains take 7–15 years to build.

That mismatch is now colliding with reality. Transformer lead times are stretching toward five years. Turbine backlogs are growing. Add rare earth constraints, and the AI buildout becomes not just expensive—but uncertain in its execution.

The Strategic Reframe

Markets still price AI as software. It is not. It is a materials-constrained industrial system—one now visibly splitting into two worlds: controlled supply inside China, constrained access outside it. Yttrium is not just a signal. It is a warning. The question is no longer who leads in AI. It is who controls the elements that determines whether AI can be built at all—and at what cost. For now, that advantage remains not only concentrated but widening.