Filed Under Healthcare and Others

The film is twenty to thirty micrometres thick, against a human hair at about eighty. The company that makes it explains the process on a page written in the register of a children's picture book: mix the raw materials in fixed proportions into a solution called a varnish, coat the varnish onto a support film, dry it, lay a protective film on top. Before this, the then-mainstream ink form of insulating material had to be applied and dried one side at a time. The film covers the copper wiring layered inside a semiconductor package substrate. The page calls it a film-shaped insulating material; the securities filing calls it an interlayer insulating material for semiconductor packages, and the page never uses that term.

That filing organises the group into three reportable segments — Seasonings and Foods, Frozen Foods, Healthcare and Others — plus Other. Electronic materials appear in the fourth tier down: inside Healthcare and Others, inside the product category Functional Materials (electronic materials and others), in a list that also contains an adhesive, a magnetic film, activated carbon and release paper. The English reportable-segment table carries the same four names. There is no semiconductor segment, no materials segment, and no financial line item called ABF — the name appears only in the column listing main products.

The numbers explain why anyone would ask. For the year ended 31 March 2026, Functional Materials revenue was ¥100,782 million (US$615 million), against ¥76,568 million (US$467 million) the year before, a rise of 31.6 per cent. Consolidated revenue was ¥1,583,719 million (US$9,669 million). So the category is 6.36 per cent of the group's revenue — a ratio the filing does not print, computed here from two figures it does.

Profit is where the proportions come apart. Healthcare and Others earned business profit of ¥66,202 million (US$404 million); the Functional Materials line within it earned ¥54,600 million (US$333 million), up from ¥40,200 million (US$245 million). Group business profit rose ¥21,800 million (US$133 million); the Functional Materials increase was ¥14,400 million (US$87.9 million), so on arithmetic from the disclosed increments the category supplied roughly two-thirds of the year's group improvement. The English factbook made the same shape visible a year earlier in pie charts: 23 per cent of the segment's revenue, 88 per cent of its profit. The company does not publish a margin for the line, and the quarterly disclosure stops at amounts — ¥28,900 million (US$176 million) of revenue and ¥16,500 million (US$101 million) of business profit in the October–December quarter.

The margin is the number that never appears next to the amounts. Divide the line's business profit by its revenue — ¥54,600 million (US$333 million) over ¥100,700 million (US$615 million), as the May segment appendix prints them, in hundreds of millions — and the year to March 2026 comes out at 54.2 per cent; the finer revenue figure the results disclosure gives for the same year, ¥100,782 million, does not move it at one decimal. On that disclosure's amounts for the year before, ¥40,200 million over ¥76,568 million, the ratio was 52.5 per cent, and the shareholder's own chart of the same line, rounded to the nearest billion yen and sourced to a blend of company disclosure, a data terminal and its own research, lands in the same place: ¥77,000 million (US$470 million) of revenue, ¥40,000 million (US$244 million) of profit, 53 per cent. A June 2026 briefing devoted to the business gives that ratio a shape and no scale: about 30 per cent for the year to March 2019, over 50 per cent for the year to March 2026, the years between unlabelled. The reason offered for the climb is composition, a rising share of newly developed and high-value-added products, and it is the only reason offered. Divide the increments and the arithmetic agrees with the reason: ¥14,400 million (US$87.9 million) of additional profit on ¥24,214 million (US$148 million) of additional revenue is an incremental margin of 59.5 per cent, above the average it lifted. All of these ratios sit above corporate common costs, which the company stopped charging to segments this year and which the appendix does not charge to the line.

The plan for the current year, published in that segment appendix in May, holds the ratio still. Functional Materials revenue of ¥111,600 million (US$681 million) and business profit of ¥60,500 million (US$369 million) work out at 54.2 per cent, last year's figure to a decimal, because the increments are thinner: ¥5,800 million (US$35.4 million) of profit on ¥10,800 million (US$65.9 million) of revenue, an incremental margin of 53.7 per cent. Growth decelerates alongside it. The company's own index puts year-on-year revenue at 125 per cent, then 131 per cent, then an initial 110 per cent against the just under 111 the appendix implies, and the briefing says the year is running ahead of that plan without publishing what ahead is. The appendix's stated reason for the increase is one clause: mainly increased sales of electronic materials.

What the increase cannot be separated into is volume and price. The briefing names two drivers, higher quantity and higher value-add, and stops; the appendix names one, higher sales; neither converts either into an amount, and currency is not mentioned anywhere in the briefing's thirty-five pages. The same company decomposes its soup stock. For Japan it publishes revenue at 111 per cent for the year on volume of 96 and unit price of 115 — and, in the spoken script rather than on the slide, 105 on volume 104 and price 101 once coffee is excluded, which turns the price story back into a volume one — then plans 106 on volume 103 and price 103. For overseas it publishes 104 on 102 and 102 for the year and plans 107 on 105 and 102, in local currency. Two indices and a caveat for the seasonings, two nouns for the film. The grain of disclosure is finest for the business that grew slowly and coarsest for the one that supplied the profit.

The one place the briefing gets specific is geometry. It charts package substrates going from three layers a side at the film's launch to, by about 2023, nine a side at up to 70 millimetres square; 2026's AI packages at eleven a side and up to 100 millimetres square; a 2031 forecast at thirteen and up to 120. Multiply layers by area and the notional footprint per package runs 18 × 4,900, then 22 × 10,000, then 26 × 14,400 square millimetres — two and a half times, then seventy per cent more. That product is not a disclosure. The dimensions are maxima rather than averages, the three points are different application classes rather than one product's history, and the briefing gives no conversion from layers and millimetres to film consumed, and no shipment quantity in any unit at all.

The 88 per cent is the number that will age fastest. It is the film line's share of its segment's business profit for the year to March 2025, charted in a shareholder's presentation this March, sourced there to company disclosure and computed before corporate common costs are allocated to segments. On the company's own published amounts for the following year the same ratio is 82.5 per cent, ¥54,600 million (US$333 million) over ¥66,202 million (US$404 million); on its plan for the current year it is 75.6 per cent, ¥60,500 million (US$369 million) over ¥80,000 million (US$488 million). Those two use published segment profit, which since this year also carries no allocated common cost; whether the deck's denominator is that denominator it does not say, so the three points are not strictly one series. Nothing shrank. The rest of the segment recovered: bio-pharma services and ingredients go from ¥9,500 million (US$58.0 million) of business profit to a planned ¥17,000 million (US$104 million). The concentration statistic everyone now quotes is the peak of a series the issuer expects to decline.

Two disclosure mechanics deserve flagging before anyone builds a series. Business profit is the company's own defined measure — revenue less cost of sales, selling, R&D and administrative expenses, plus equity-method income — not an IFRS line, and it excludes the other operating income that carried group operating profit to ¥199,412 million (US$1,217 million) this year on a fixed-asset disposal gain of ¥41,200 million (US$252 million). Separately, corporate common costs stopped being allocated to segments this year, with the prior year restated; segment profits published before that change are not continuous with these.

The ten-year data book has the subtler hole. Its segment table prints four business profits that sum to ¥223,759 million (US$1,366 million) against a stated total of ¥181,163 million (US$1,106 million); the ¥42,596 million (US$260 million) between them is consistent with the corporate common cost that stopped being allocated, though the table carries no row for it and the amount would have to be confirmed against the segment reconciliation in the securities report. The same quantity is printed in the notice of the June shareholder meeting, as minus ¥42,500 million (US$259 million), the difference truncated to hundreds of millions. The prior year reconciles the same way, by derivation rather than by print: the notice gives the change as minus ¥2,700 million (US$16 million), which puts the year before at minus ¥39,800 million (US$243 million), and in the data book the four segments overshoot that year's total by ¥39,861 million (US$243 million). The book's depreciation, capital expenditure and research tables all carry a company-wide line. The profit table, alone, does not.

What is not disclosed is the film itself. The category's profit is published — ¥54,600 million (US$333 million) for the year, in a segment appendix — but Functional Materials is a composite: an adhesive, a magnetic film, a resin-coated copper foil and a sealing material are all inside it, and revenue or profit for ABF alone appears in neither the forecast tables nor the seventeen-page data book, while the securities filing carries the name only in a list of main products. The data book does not contain the words Functional Materials at all. A reader who wants the economics of the insulating film has the economics of a basket that contains it.

The asymmetry is sharper in English. The factbook gives Functional Materials its own page, with an eight-year chart of revenue, business profit and margin — and no data labels, no table, the chart embedded as an image. The multi-year segment exhibits are all built that way. The neighbouring Bio-Pharma line gets a revenue figure of ¥147,600 million (US$901 million) for the year to March 2025, a 40/60 internal split, and a multi-sentence account of its history, leading share and CDMO scope — though inside the Healthcare and Others overview rather than on a page of its own. Functional Materials gets three sentences and a reference link to a Japanese-only investor briefing. Nothing here is undisclosed, exactly. It is disclosed in a format that does not survive translation.

That Japanese-only briefing, when it came on the last day of June, turned out to be the strangest document in the file: thirty-five pages on the business, and the business's revenue and profit appear nowhere in them as a yen amount. Both are index bars. The revenue chart's vertical axis carries the labels 0 through 1800 in the PDF's text layer, not visible in the page as rendered and attached to no unit; its final point, the one for 2030, is a bar and an arrow. A later outlook chart for 2030 has no axis at all, three bars. The horizontal axis is broken twice. Year-on-year percentages are printed for three years and omitted for two, one of which the shareholder's own estimates put at a decline. The appendix in May had printed the same business as two amounts and a clause. A briefing dedicated to the business discloses less arithmetic than the routine appendix that forecasts it.

The share claim arrives in two registers from two authors, and neither of them measures anything. The June briefing says over 95 per cent, maintained since the 1999 launch, and names no measuring body. The shareholder's deck says about 95 per cent on page four, over 95 per cent on page fourteen, and about 100 per cent for high-end AI data-centre chips; page four is also where the product is called a de facto standard with no qualified alternative at the high end. Neither document says whether the share is by value or by area, over what market boundary, or for which year, and the deck's footnote on those pages blends company disclosure with a data terminal and its own research. The share is the one number the two authors agree on, to within a word.

The shareholder arrived on 31 March, by newswire from London rather than by regulatory filing. Palliser Capital, describing itself as a top-25 holder, published twenty-eight pages with three asks: make the film its own reportable segment, raise its price by more than 30 per cent, and restructure frozen foods to a return on invested capital above 8 per cent. There is no buyback request, which is what the same manager has asked of other Japanese issuers, and no list of what the new segment would have to publish — no volumes, no average price, no utilisation, no order book. The footnotes are the instructive part: exactly three pages are sourced to company disclosure alone, one page to the manager's own analysis alone, and the rest to combinations of company disclosure, a data terminal, broker research, interviews with industry specialists and that same analysis. The only broker named anywhere is Morgan Stanley, on a page about other companies' price increases. The disclaimer says the document rests on public information and the manager's analysis of it, relied on without independent verification and warranted as to nothing. The deck also argues valuation, which is not reproduced here.

The price ask rests on a chain of small ratios: the film is about 10 per cent of the bill of materials of the substrate, the substrate under 1 per cent of the selling price of an AI accelerator, the film itself under 0.1 per cent. Those come from broker research the deck neither names nor dates. Five weeks later, the company's results deck gave the film's pricing a page, and said that sustained increases in its unit price had been achieved as higher-priced products came to occupy a larger share of the mix, performance rising with them — performance and price, it wrote, are inseparable. It identifies no counterparty. Whether it is answering anything is not determinable from the document.

The film is made by Ajinomoto Fine-Techno, a subsidiary founded in September 1942 with capital of ¥315 million (US$1.92 million) and 487 people as of 1 April 2026. What it makes is narrower than the film. The briefing says the company specialises in producing the varnish, at Kawasaki and Gunma, and outsources the coating, the slitting and the warehousing: everything after the mixing. It does not name the contractors, count them, or give their share of the work. So the capacity question cannot be closed from the company's own capital spending: the varnish is the part it builds, and the line that turns varnish into film sits on a balance sheet the briefing addresses with the sentence that the supply chain as a whole presents no concern.

Capital spending is where the reader's inference has to stop. Healthcare and Others invested ¥27,833 million (US$170 million) in the year to March 2026, on the data book's segment table, which gives the amount and nothing else. The securities report for the year to March 2025 — when the segment's capital expenditure was ¥32,267 million (US$197 million) — lists one material new-construction plan for it, an amino-acid capacity expansion in the Americas at ¥3,550 million (US$21.7 million), and no electronic-materials project at all. The film's numbers sit outside the filing: a Gunma plant for advanced varnish at about ¥10,000 million (US$61.1 million), running since 2025, inside an investment frame of ¥25,000 million (US$153 million) for 2023 to 2030 — the same figure the company attached to increased film production in its February 2023 roadmap, though the briefing calls it a frame and not committed spending — and land in Gifu for a third site, announced in May 2026, investment amount undisclosed, capacity given only as comparable to Gunma's, operating from 2032. Comparable to Gunma is the only quantity given. Capacity is never quantified in any unit; utilisation, depreciation and payback do not appear in the briefing at all, and the previous year's securities report does not contain the word Gifu.

Two things the briefing never says. It does not mention a competitor, a glass core or a photoimageable insulator in any of its pages, and it frames technology change as extension rather than defence — the same technology carried into resin-coated foil, redistribution layers, sealing materials and, past 2030, optical waveguides. The securities report for the year to March 2025 carries no major-customer note and no by-customer sales table either, which is arithmetic rather than comfort: the threshold is 10 per cent of consolidated revenue, ¥153,056 million (US$934 million) that year, twice what the results disclosure gives as the revenue of the entire Functional Materials category in the same year. A single buyer could take every metre the company sells and still never have to be named.

So the filing holds both things at once and does not reconcile them. A Seasonings and Foods segment of ¥936,926 million (US$5,720 million) in external revenue, of which seasonings proper are ¥478,038 million (US$2,918 million), descended from a patent granted in July 1908 and a product on general sale from May 1909 — and an interlayer insulating film for semiconductor packages, the driver the company names first for the Healthcare segment's ¥20,500 million (US$125 million) increase in business profit, though product-category profit is not disclosed. One set of notes, one common-cost allocation policy that changed this year. The English reader gets the film as an unlabelled chart. The Japanese reader gets a product-category revenue line and no product-category profit. Everybody gets the film.

On currency: yen amounts are converted at ¥163.8 = US$1, the rate on 24 July 2026, applied uniformly to every period in this piece regardless of the period's own exchange rates; the dollar figures are therefore a unit of account, not a translation of the company's own results, which the company reports in yen and which include translation effects at rates it discloses separately. Significant figures in the dollar amounts follow the yen figure as originally published. Ratios, percentages, multiples, headcounts and site counts are unconverted. Amounts published in oku-yen have been read as hundreds of millions of yen before conversion.