NextFin News - Fujifilm Holdings is scaling production of the photoresists and polishing slurries that advanced chips cannot be made without, betting that the AI buildout will keep its semiconductor materials business growing toward a 500 billion yen sales target by 2030. The move puts the Japanese supplier at the center of the most concentrated supply chain in the chip industry - where a handful of firms control the chemistry that determines whether a wafer works at all.
The expansion, announced in Tokyo on Friday, extends a capital-spending program that has already committed more than 100 billion yen to semiconductor materials facilities since fiscal 2021, with another 100 billion yen planned through fiscal 2026. It lands as Fujifilm's electronic materials unit reported 29.3 percent revenue growth in the fiscal year ended March 2026, driven by demand for advanced materials used in generative AI.
The Situation: A Quiet Supplier Riding the AI Wave
Fujifilm is not a household name in artificial intelligence. It does not make the GPUs that train large language models, the high-bandwidth memory stacked beside them, or the lithography machines that print circuit patterns onto silicon. It makes the photoresist - the light-sensitive chemical film that carries those patterns - and the chemical-mechanical polishing slurries that flatten each layer before the next one is printed. Without both, no leading-edge chip leaves a fab.
That positioning has turned Fujifilm's electronics business into one of the group's fastest-growing units. In the fiscal year ended March 31, 2026, the Electronics segment posted revenue of 127.5 billion yen, up 27.4 percent year over year, with operating income jumping 74.0 percent to 30.7 billion yen. Within it, the Electronic Materials business grew 29.3 percent. For the full year, the group recorded consolidated revenue of 3.357 trillion yen, up 5.0 percent, and operating income of 350.2 billion yen, up 6.1 percent - record highs across revenue, operating income, and net income, according to the company's May 12 results release.
President and CEO Teiichi Goto attributed the record directly to the chip-materials upswing. "We achieved record-high revenue, operating income, and net income, driven by strong performance in the Healthcare segment supported by the operation of new Bio CDMO facilities, solid results in semiconductor materials within the Electronics segment, and steady growth in the Imaging segment, including digital cameras," Goto said.
The growth trajectory is built into the company's medium-term plan. Fujifilm's semiconductor materials sales grew at an annual rate of roughly 20 percent between fiscal 2021 and fiscal 2024, and the company has set a fiscal 2030 target of 500 billion yen in semiconductor materials sales - approximately double the fiscal 2024 level. The broader advanced semiconductor market is expected to double by 2030, the company said in a November 2025 announcement.
That ambition requires capacity, and capacity is what Fujifilm is adding. The company has invested more than 100 billion yen in semiconductor materials facilities from fiscal 2021 to fiscal 2024 and plans to invest an additional 100 billion yen or more from fiscal 2025 to fiscal 2026. The spending is geographically deliberate: a 13 billion yen building at the Shizuoka site in Japan for advanced resists including extreme ultraviolet (EUV) photoresists, which began operations in the fall of 2025; a 7 billion yen expansion at the Oita site for post-CMP cleaners that lifts that plant's capacity by about 40 percent, due in the spring of 2026; a roughly 4 billion yen investment at the Zwijndrecht site in Belgium for CMP slurries and photolithography developers, also starting up in the spring of 2026; and new equipment at facilities in Pyeongtaek, with production coming online in the autumn, and Cheonan, where an abrasive-agent plant is set to raise output capacity by 30 percent once mass production begins in the spring of 2027.
The question the market is asking is whether this is a cyclical inventory upswing dressed up as secular growth - or a genuine structural shift in where chip-industry profits sit.
Why Photoresist Is the Choke Point
To understand why Fujifilm's expansion matters, start with the mechanism. A modern logic chip is built in hundreds of steps, and photolithography - the process of projecting circuit patterns through lenses onto a wafer coated in photoresist - is repeated dozens of times. At the most advanced nodes, each layer is measured in single-digit nanometers. The photoresist must resolve features smaller than the wavelength of the light used to expose it; EUV lithography, which uses 13.5-nanometer light, demands resists that can pattern reliably at that scale while surviving the intense radiation.
Only a few companies in the world can make EUV photoresist at the purity and consistency required for high-volume manufacturing. The top three players - Tokyo Ohka Kogyo, JSR, and Shin-Etsu Chemical - hold more than 90 percent of the EUV photoresist market, according to market research. Fujifilm is one of the challengers, having launched its negative-tone EUV resist and EUV developer in October 2024 and completed a new development and evaluation building at Shizuoka in November 2025.
This concentration is not an accident of history. It is the product of a qualification barrier that makes the photoresist business structurally different from most of the semiconductor supply chain. A chipmaker cannot simply switch resist suppliers between production runs. Changing the chemistry requires requalifying the entire lithography process - a cycle that can take a year or more and carries the risk of yield loss on millions of dollars of wafers. Once a resist is qualified into a process, it tends to stay there. That gives incumbent suppliers pricing power and revenue visibility that equipment makers, whose sales are lumpy and cyclical, do not enjoy.
There is a second mechanism at work: the AI workload is changing what gets built. AI accelerators are large-die, advanced-node chips that consume disproportionate amounts of leading-edge capacity. A single AI GPU can occupy the wafer area of many conventional processors, and each additional metal layer means more lithography steps, more resist, and more polishing cycles. High-bandwidth memory stacks, essential for AI systems, add further process steps. The result is that demand for advanced materials grows faster than the unit count of chips - a multiplier effect that benefits suppliers of the chemistry at the leading edge more than suppliers of commodity materials.
The market numbers bear this out. The global EUV photoresist market is projected to grow from US$226.4 million in 2024 to US$878.9 million by 2030, a compound annual growth rate of 25.4 percent, according to market research. The broader global chipmaking materials market is forecast to expand 35 percent to US$58.3 billion by 2029 compared with 2023 levels, according to research by Fuji Keizai. CMP slurries, another Fujifilm strength where the company says it holds world-leading share, are expected to grow at an annual rate of 13 percent.
The Cyclical Question: Is This a Wave or a Tide?
Here is the judgment that separates a trade from an investment thesis. Semiconductor materials are, at their core, cyclical. Chip demand rises and falls with inventory cycles, consumer electronics seasons, and macroeconomic conditions. The fiscal 2023 downturn saw chipmakers cut utilization and draw down materials inventories; the 2025-2026 upswing is, in part, the rebound from that trough. If the AI buildout slows - if hyperscalers digest their GPU fleets and delay the next data center generation - materials orders will fall with them. A cyclical reading says Fujifilm's current growth rate is a mean-reverting wave, and the 20 percent annual growth of fiscal 2021-2024 cannot be extrapolated indefinitely.
But there is a structural argument, and it is the stronger one for the leading-edge materials that Fujifilm is expanding. Three pieces of evidence support it.
First, the demand driver is not a single product cycle but a platform shift. The transition from general-purpose computing to AI-accelerated computing is rewiring capital expenditure across the entire technology sector. Hyperscalers and large enterprises are committing to multi-year data center expansion plans that extend beyond any single chip generation. This is not the smartphone cycle, where a hit product drives a two-year surge and then saturates.
Second, the supply constraint is physical and slow to relieve. New chemical production capacity takes years to permit, build, and qualify. Fujifilm's own timeline - announcing investments in 2024 and 2025 for facilities coming online in 2025 and 2026 - illustrates the lag. In a market where qualified suppliers are few and capacity additions are slow, the pricing power of incumbents persists through the cycle, not just at its peak.
Third, the qualification barrier I described above converts cyclical volume into sticky revenue. Even in a downturn, chipmakers are reluctant to requalify a cheaper alternative when the cost of a yield excursion dwarfs the savings. This asymmetry - easy to enter, hard to exit - is the hallmark of a structural advantage.
The balanced read: the volume component of Fujifilm's growth is cyclical and will fluctuate with the chip cycle, but the mix component - the shift toward advanced-node materials with higher content-per-wafer and higher barriers - is structural. Fujifilm's expansion is a bet that the structural shift in mix will outlast the cyclical swings in volume. That is a defensible position, but it is not risk-free.
The Counter-Thesis: Concentration Cuts Both Ways
The strongest argument against Fujifilm's expansion is that the same concentration that gives suppliers pricing power also gives customers leverage - and that leverage is about to increase. TSMC, Samsung, and Intel are the only buyers of EUV photoresist at scale, and each is actively pursuing second sources and, in some cases, in-house development to reduce dependence on Japanese chemical suppliers. Samsung has long cultivated relationships with domestic and Korean suppliers; China, locked out of EUV lithography tools by export controls, is pouring state capital into developing its own photoresist and materials base.
There is also a margin argument. Fujifilm's first quarter of fiscal 2027 - the three months ended June 30, 2026 - showed revenue of 826.5 billion yen, up 10.3 percent year over year, but operating income of 51.2 billion yen, down 32.0 percent. The company said the profit decline was broadly in line with its internal plan, but the divergence between top-line growth and bottom-line contraction is a warning sign. Rising raw material and energy costs, the ramp-up expenses of new facilities, and the front-loaded nature of capital investment all compress margins before new capacity pays for itself. If the AI-driven revenue growth slows before the new plants reach full utilization, Fujifilm could face the worst of both worlds: cyclical volume pressure on top of structural margin compression from overcapacity.
Finally, valuation discipline matters. Market data showed Fujifilm's shares fall approximately 17.9 percent the session after its August 6 earnings release, a sharp reminder that the market penalizes profit deterioration even when revenue is growing. The stock has since recovered some ground; as of the Tokyo session on August 21, it was trading in the low 3,200 yen range with a market capitalization of about 3.98 trillion yen, still well below its 52-week high near 3,908 yen.
These counter-arguments are substantial, but they do not overturn the structural thesis. Customer concentration is real, yet the qualification barrier limits how quickly buyers can act on it. Margin pressure is real, yet it is the predictable cost of a capacity build-out that the company has explicitly guided toward. The risk is not that Fujifilm is wrong about the direction of the market; it is that the timing of its capital deployment could leave it exposed if the AI capex cycle turns sooner than expected.
What to Watch
The falsifying signal for the structural thesis is specific: if Fujifilm's Electronic Materials revenue growth decelerates to single digits for two consecutive quarters while the company's guided capital expenditure remains elevated, the structural-mix argument weakens materially and the cyclical reading takes over. Watch the quarterly Electronics segment revenue and operating income figures, the utilization disclosures for the Shizuoka and Oita facilities, and any change to the 500 billion yen fiscal 2030 sales target.
By time horizon, the picture splits. In the short term, sentiment will track the AI capex cycle and quarterly margin prints - the August 6 reaction shows how quickly the market can turn on profit deterioration. Over the medium term, the key variable is whether the new capacity comes online on schedule and achieves the utilization needed to restore operating leverage. Over the long term, the question is whether Fujifilm can convert its EUV and advanced-packaging materials into qualified, sticky positions that survive the next downturn.
Three scenarios frame the outlook. The base case: AI-driven advanced-node demand continues to grow in the high teens to low twenties annually, the new facilities ramp on schedule, and Fujifilm moves steadily toward its 500 billion yen target. The upside case: EUV adoption accelerates with high-NA lithography, Fujifilm wins meaningful share from the top-three incumbents, and the materials mix shifts even further toward the leading edge. The downside case: hyperscaler capex slows in 2027, memory prices remain elevated and pressure customer budgets, and new capacity comes online into a softer market, compressing margins.
"In this current fiscal year, our focus is on achieving sustainable growth by expanding our semiconductor materials business and scaling production capacity in the Bio CDMO business," Goto said in May. "We are committed to accelerating progress toward the goals outlined in VISION2030 and establishing Fujifilm as a collection of global leading businesses."
The company has also said it "will continue to expand production capacity and develop technologies to meet diverse customer needs, including increasing demand for AI semiconductors," and that it sees "the advanced and next-generation semiconductor market, which is expected to double by 2030."
The Bottom Line
Fujifilm's expansion is a calculated bet that the AI revolution will be measured in chemistry as much as in silicon. The company is not trying to win the GPU race; it is trying to own the materials without which the race cannot be run. That is a narrower ambition, but also a more defensible one - protected by qualification barriers, reinforced by a capacity lag that favors incumbents, and aimed at a market the company expects to double within five years.
The risk is timing, not direction. If AI capital expenditure proves cyclical rather than structural, Fujifilm will have built capacity it does not need at margins it cannot sustain. But if the current buildout is the first inning of a decade-long platform shift, the company is buying optionality on the most valuable real estate in the chip supply chain at a price the market has not fully recognized.
For investors, the takeaway is not that Fujifilm is a pure AI play. It is that the AI trade has a long tail of suppliers, and the chemical layer - quiet, concentrated, and hard to replicate - may capture more durable value than the flashier names that dominate the headlines. The market's verdict will come not from the announcement, but from the quarterly utilization and margin data that follow.
Fujifilm's expansion is not a bet that AI will win. It is a bet that whoever wins AI will still need Fujifilm's chemistry - and that is a wager that pays off in more scenarios than it fails.
Explore more exclusive insights at nextfin.ai.
