Why Japan Bought Its Photoresist Champion

In 2023 and 2024, the Japanese government bought a chemical company for about $6.4 billion. It did not make chips, tools or wafers. It made a light-sensitive film that gets spun onto a wafer. The company was JSR, the buyer was the Japan Investment Corporation, a government-backed vehicle, and the stated purpose was to protect Japan's national security interest in semiconductor chemicals.

That transaction is the best single piece of evidence I know for how the industry itself values one small category of specialty chemistry. In this piece I want to explain what a photoresist does, why supply of the advanced grades is so concentrated, what the other chemical chokepoints look like, and how the risk fits into the wider picture of single points of failure in electronics manufacturing. The figures come from the semiconductor materials and supply-chain risk chapters of my ESDM research book, and where the book's own figures differ slightly I say so.

What a photoresist actually does

A finished, polished wafer is a substrate, not a chip. To pattern it, a fab needs a second category of materials: specialty electronic chemicals. These include advanced photoresists, chemical-mechanical polishing slurries, high-purity wet-etching acids and specialty process gases. They are the working fluid of every lithography, etch, deposition and cleaning step.

Photoresists are the most consequential of these. A photoresist is a light-sensitive film spin-coated onto the wafer. It is exposed through a lithography mask and developed, and what remains defines the pattern that gets etched into the layer below. The exposure wavelength sets the resist chemistry. Two grades matter most at the leading edge: extreme ultraviolet at 13.5 nanometres, and 193 nanometre immersion, usually shortened to ArFi.

The reason this is a chokepoint and not just a consumable is that a resist has to be formulated for a specific exposure tool and process, and the leading-edge chemistries are the product of long development effort. A fab cannot swap one in on short notice.

The concentration numbers

JSR Corporation controls more than 30% of global EUV and ArFi photoresist supply on its own. Japan as a whole, through JSR, Tokyo Ohka Kogyo, Shin-Etsu, alongside the US-headquartered DuPont, accounts for more than 85% of global EUV photoresist supply. A related figure in my supply-chain risk chapter, taken from a BCG and SIA analysis, says more than 80% of advanced photoresists originate in Japan. The two numbers are different measures with different scopes, so I would not reconcile them into one. Either way, the direction is the same.

At the most advanced end, JSR acquired Inpria, which gave it a proprietary position in metal oxide resists. These are tin-based EUV resist chemistries developed for sub-2nm pitch scaling. The book describes this as functioning less like a market-share statistic and more like a patent-protected monopoly on a chemistry class for which the rest of the industry does not yet have an alternative. That is a strong claim and I would flag it as the book's characterisation, not a measured share.

Why Tokyo bought its own champion

The buyout price was ¥909 billion, roughly $6.4 billion. It is a large sum for a firm that does not manufacture a single transistor. The logic follows from the concentration. If most of the world's advanced photoresist supply runs through three or four Japanese chemical formulators, then Japan's leverage over the entire downstream logic and memory industry rests on private companies whose ownership, investment decisions and export posture could otherwise drift outside government control, through an acquisition, geopolitical pressure or a change in corporate strategy. The state purchase turns a market position that already existed into one the government can defend directly.

Other chemical chokepoints sit right next to resists

Photoresists are the loudest case, but the book documents at least two further categories with comparably tight concentration, plus a third where it declines to guess.

CMP slurries are colloidal silica or cerium-oxide abrasive-chemical mixtures used between processing steps. Together, CMC Materials (a subsidiary of Entegris), Fujimi and Resonac hold more than 70% of the US and Japan-concentrated market. High-purity hydrofluoric acid, refined to 9N for oxide etching and cleaning, is supplied by three companies at above 80% Japanese market concentration. For specialty gases such as nitrogen trifluoride, tungsten hexafluoride and silane, the book says public disclosure of supplier-level concentration is thinner and treats it as an evidence gap instead of estimating.

The purity bar is a barrier of its own

Across these categories, purity standards are severe. SEMI C-series chemical standards specify sub-parts-per-trillion impurity levels for the most advanced-node chemistries. That is roughly a thousand times tighter than the parts-per-billion standard that governs the purified trichlorosilane feeding polysilicon. This is one reason wafer materials sit closer to specialty chemistry than to bulk processing. A supplier has to prove purity and lot-to-lot consistency for every customer, which is another form of qualification lock-in.

How this fits the three kinds of chokepoint

My supply-chain risk chapter sorts single points of failure into three structural categories, and photoresists touch two of them.

Two features of the geographic category matter for photoresists. First, none of the concentration figures describes a single-company monopoly. They describe a national cluster that a disruption anywhere within it would degrade at once. Second, seismic and physical exposure is a shared feature of the region. My chapter notes that Japan's and Taiwan's main fab clusters sit within active seismic fault zones, and a magnitude 6.0 earthquake is enough to trigger quartz furnace tube breakage and tool recalibration across affected fabs. That is about fabs, not chemical plants, but it shows how one region's exposure compounds.

A regulatory wrinkle: PFAS

There is also a regulatory pressure that touches resist chemistry. The European Chemicals Agency's proposed universal restriction on PFAS targets more than 10,000 fluorinated chemicals, including PTFE and other fluoropolymers used across photoresists and process coolants. It has no national-security motive, but the practical effect is similar: it forces reformulation or requalification of chemistries the industry has relied on for decades. Given how long resist qualification takes, that is not a light burden. I do not have figures on the scale of the reformulation effort, so I would treat it as a risk to watch, not a quantified one.

Why a newcomer cannot buy its way in

Every stage of the semiconductor materials chain shares the same structural feature: capital intensity compounded by know-how that does not transfer. The book puts it this way for polysilicon refining, and it applies to resists equally. A resist maker is competing against decades of accumulated formulation learning, embedded in tools and in customer processes. On top of that comes qualification with each foundry. A new resist entrant would need to prove itself in an EUV process where a bad batch can scrap wafers already carrying months of processing.

For India, the book is explicit that wafer substrate manufacturing and advanced specialty-chemical production are capabilities not yet established domestically, and it marks a concrete roadmap as an evidence gap. Its point is that a critical-minerals mission does not by itself create a photoresist formulator able to compete with JSR's patent-protected chemistry. Those are separate gaps, and in some ways harder ones, because they resemble decades of non-transferable process learning more than a mineral-sourcing agreement negotiated on a multi-year timeline.

Reading the JIC deal as a signal

It helps to think about what the buyout does and does not tell us. It does not tell us JSR was failing. It tells us that the Japanese government judged a private company's position to be too important to leave to ordinary ownership dynamics. The buyer, the Japan Investment Corporation, is government-backed, and the stated logic was national security interest in semiconductor chemicals. In my reading, the transaction converts an accident of industrial history into a defended asset. Markets rarely do that on their own.

A second point follows. The price, about $6.4 billion, is not being paid for revenue from making chips. It is being paid for optionality over who can run leading-edge lithography. If most advanced resist supply runs through three or four formulators, the owner of one of them has a voice in downstream logic and memory production that no purchase contract gives. The state now holds that voice directly.

The chemistry that makes resists hard to replace

To see why substitution is difficult, consider what the Inpria acquisition gave JSR. Metal oxide resists are tin-based EUV chemistries developed for sub-2nm pitch scaling. The book characterises JSR's position as patent-protected, on a chemistry class for which the rest of the industry does not yet have an alternative. I cannot verify the patent landscape myself, so I read that as the book's characterisation. But the structure is credible: at the tightest pitches, a resist is not a generic input. It is tied to the exposure wavelength, the tool and the etch process that follows it.

A consumable that is not a commodity

Photoresist is consumed continuously, wafer after wafer, so it behaves like a commodity in the accounting sense. In the process sense it is closer to a component. A fab qualifies a resist for a given layer, and changing it means requalifying the layer. That combination, continuous consumption and hard substitution, is what makes a supply interruption dangerous: the fab cannot pause and buy something else the next day.

The parallel with ABF resin

The same structure appears in another chemical my source names. Ajinomoto Fine-Techno supplies roughly 100% of global ABF resin film, the build-up material for advanced package substrates. The failure scenario in my source is blunt: a chemical-plant accident halts global production of the substrates used by server CPUs and GPUs. That is a single-supplier monopoly, the first of the three chokepoint categories. Photoresists sit between categories one and two. JSR alone is not a monopoly, but the Japanese cluster is a geographic concentration, and JSR's newest chemistry approaches the single-supplier case.

The rare-earth link that most people miss

One thread connects the chemical layer to the mineral layer beneath it. Cerium oxide is among the abrasive chemistries used in CMP slurries, and cerium is a rare earth. So a slurry market that my source shows dominated by CMC Materials, Fujimi and Resonac sits on top of a rare earth supply that China refines at 85% to as high as 99.9% by capacity estimates. In other words, the chemical layer is not independent of the mineral layer. Each step up the chain inherits the concentration of the step below it. I take this as a reason not to read any single concentration figure in isolation.

The same logic applies to photoresists, though my source does not trace individual resist feedstocks, so I do not extend it. What I can say is that a specialty chemical maker, however dominant, still sits inside a supply chain it does not fully control.

What mitigation looks like for a chemical chokepoint

My supply-chain risk chapter makes a general point that I find useful: each of the three chokepoint categories calls for a different mitigation. For chemicals that means thinking separately about three levers. For the single-supplier case, the lever is qualifying a second source in advance, which is slow and costly precisely because of the qualification barrier. For the geographic case, the lever is spreading production across regions, which for resists means building formulation and manufacturing capability outside Japan, a multi-year effort. For the policy case, the lever is understanding the rules, and here the EU PFAS proposal is instructive. It is not aimed at anyone, but it forces reformulation or requalification of fluoropolymer chemistries used in resists and coolants.

What I would not claim is that any of these levers is quick. The book's own evidence is that qualification and process learning are the slow variables. A government that wants resilience in specialty chemicals has to start earlier than it would for equipment, and it has to accept that the first product may be a partial substitute, not a full one.

India and the chemical gap

India does not appear in the photoresist market in my source, and I would not invent a role for it. The relevant point is a negative one: advanced specialty-chemical production is a capability India has not yet established domestically, and my source marks any concrete roadmap as an evidence gap. That matters for the current wave of Indian fab and packaging investment. A fab located in India would import its resists, its slurries and its ultra-pure acids, and would face the same concentration as every other fab, with less local buffer.

The book draws a distinction I find sharp. Securing minerals, wafers and chemicals are three separate problems, each with its own qualification timeline and its own incumbents. Winning the mineral problem does not solve the chemistry problem. A country building from a low base is only as resilient as the least secure of the three layers beneath its fab.

What to take away

I would leave you with four points. One, photoresists are formulated products tied to a specific exposure process, which is why they are hard to substitute. Two, JSR holds more than 30% of EUV and ArFi supply alone, Japan holds more than 80% to 85% depending on the scope measured, and the Inpria acquisition gave JSR a tin-based EUV chemistry the book treats as close to a patent monopoly. Three, the 2023-24 state buyout for about $6.4 billion is the clearest signal that governments now read specialty chemistry as a security asset. Four, the chemical layer sits beside other tight concentrations, in CMP slurries and ultra-pure hydrofluoric acid, and beside gaps in the public data on gases that I would not fill with guesses.