Chapter 2.10 — Materials & Inputs

At the very bottom of the stack sit the raw and refined materials that everything else is built from, and this is the quiet chokepoint that most investors underweight. A ten-dollar consumable can gate a forty-thousand-dollar GPU. The layer's defining feature is a mutual hostage situation: China controls the raw minerals the West cannot easily replace, and Japan and the West control the ultrapure engineered materials China cannot yet make. Each side can hurt the other, which is why this layer is less an investment theme than a map of where the real leverage in the whole US-China contest actually lies.

Before a chip is designed, fabricated, or packaged, it begins as materials: the silicon wafer it is printed on, the light-sensitive photoresist that patterns it, the ultrapure gases that etch it, and, further upstream, the rare-earth elements and specialty minerals woven through the magnets, lasers, and components of the whole system. These inputs are cheap relative to the finished chip, which is exactly why their leverage is so disproportionate: a shortage of a single obscure material can halt a production line worth billions. The layer splits into two worlds with opposite balances of power, and understanding that split is the point of this chapter.

China's half: the minerals

China's dominance of the raw-mineral base is close to total in several categories, and it has shown, repeatedly, that it will use it.

China controls roughly 90% of the world's rare-earth processing, about 98% of gallium and 77% of germanium refining, more than 95% of the graphite anode material for batteries, and around 90% of high-performance magnet production. Over 2024 and 2025 it turned these into weapons in sequence, banning gallium, germanium, and antimony exports to the US, then adding heavy rare earths and magnets, then a rule extending control to any foreign product containing even trace Chinese-origin rare earths. The effect was violent where it landed: antimony, used in everything from munitions to semiconductors, ran from about $1,400 a tonne to nearly $60,000, a rise of more than 2,600%, before a truce eased it. That truce, agreed after the Trump-Xi meeting in late 2025, suspends the harshest controls only until late 2026, so the leverage is paused, not removed.¹

Critical Materials and Component Suppliers

The materials layer runs across categories, and the balance of power flips from China to Japan and the West as you move from raw minerals to engineered inputs.

The West and Japan's half: the engineered materials

Run the same analysis on the ultrapure, engineered materials, and the dominance flips to the other side of the Pacific.

Japan makes roughly 90% of the world's semiconductor photoresist and more than 95% of the high-end EUV resist, through Shin-Etsu, JSR, and Tokyo Ohka; JSR's ownership of Inpria, the leading metal-oxide EUV-resist maker, tightens that grip further. Shin-Etsu and SUMCO together supply more than half of all 300mm silicon wafers, with GlobalWafers, Siltronic, and SK Siltron rounding out an oligopoly that controls over 80% of the market. The ultrapure gases that etch and deposit run through Air Liquide, Linde, and Air Products, with a lingering vulnerability in neon (much of it historically from Ukraine, which spiked prices ninefold after the 2022 invasion). The CMP slurries and wet chemicals that polish and clean are led by America's Entegris and Japan's Fujimi and Resonac. And, as Chapter 2.7 detailed, Japan's Ibiden and Shinko hold more than 70% of high-end substrates while a single company, Ajinomoto, makes more than 95% of the insulating film. The materials that require the most process sophistication are held almost entirely by the US, Japan, and Europe, and China cannot yet substitute them at scale.²³

This is the mirror image of the minerals story, and together they define the layer. China holds the upstream raw base; the allied bloc holds the downstream engineered tier. Each can inflict real damage on the other, which produces a fragile, mutually-deterred stability and a scramble on both sides to build redundant supply, the "de-Americanization" of China's bill of materials matched by the West's "de-risking" from China. Both efforts are expensive, slow, and incomplete.

Qualification economics: small cost, large switching friction

The investment logic in engineered materials differs from ordinary commodities. The material may be a small fraction of finished-chip cost, but a purity deviation can destroy yield across an expensive production line. Customers therefore qualify a supplier and a specific formulation through extended process testing; once qualified, they are reluctant to change merely to save a small amount on input cost. That creates retention and pricing power for proven suppliers, while making new capacity slow to convert into commercial share.

Scarcity can still be temporary. Inventory buffers, dual sourcing, recycling and customer redesign gradually weaken a shortage, and commodity prices can fall before a new mine or refinery reaches steady production. The relevant dashboard is consequently broader than spot price: qualification wins, long-term offtake, customer inventories, production yield, cash cost, government support and the time remaining on export controls. Ajinomoto's position in build-up film illustrates the difference between a qualified process dependency and a generic raw material.³

For rare-earth projects, strategic value and minority-shareholder returns can diverge. A government may rationally support redundant domestic supply even when the project's through-cycle economics are poor. Public investors must distinguish policy support that protects capacity from contracts, price floors or capital structures that actually protect equity returns.

Truce expiry, qualification wins and substitution

The principal near-term policy variable is the minerals truce, which expires in late 2026; whether China re-tightens rare-earth and gallium controls, or lets the truce hold, is a genuine swing factor for defense, magnet, and semiconductor supply chains. Watch whether Western rare-earth capacity, MP Materials and Lynas above all, scales fast enough to matter or stays a rounding error against Chinese output. Watch the engineered-materials chokepoints for their own squeezes, the Ajinomoto price increase for the second half of 2026 being a small example of pricing power at a true single-supplier bottleneck. And watch the slow migration to glass substrates, which could reshuffle the substrate hierarchy over the second half of the decade.

From chokepoint to shareholder return

The most direct public expressions of this layer are the ex-China rare-earth builders, MP Materials (MP) and Lynas (LYC.AX), which are as much policy bets as commodity plays, backed by government price floors and offtake because their strategic value exceeds their current economics. The engineered-materials champions, Shin-Etsu, SUMCO, JSR, Tokyo Ohka, Ibiden, and Ajinomoto, are concentrated process dependencies, though most trade in Japan and may be less accessible to a US investor; Entegris (ENTG) is a listed American materials supplier, and Air Liquide and Linde (LIN) provide diversified industrial-gas exposure. The layer is most useful as a map of bargaining power; the decision layer separately tests whether that strategic leverage is material to each security and already reflected in valuation.

When the leverage gets used

The mutual-deterrence framing breaks if one side decides the leverage is worth using despite the cost. A Chinese decision to let the truce lapse and re-impose hard mineral controls would spike prices and disrupt supply chains well beyond semiconductors, a tail risk that is paused rather than gone. In the other direction, the thesis for the Western rare-earth names weakens if Chinese supply floods back and collapses prices, which has ended every previous attempt to build a non-Chinese rare-earth industry; the government price floors under MP Materials exist precisely because this has happened before. And the allied advantage in engineered materials, seemingly secure, would erode over years if China's heavy investment in domestic wafers, photoresist, and substrates finally began to close the quality gap, which it has not yet.

For the downstream package, rack and 100 MW facility boundaries that consume these inputs, see §2.13.

Sources

Linked evidence for this chapter's figures and load-bearing claims: ^{2 1 3}