NGK Insulators / NGK Corporation: The Ceramics Company Betting Its Future on AI Chips Instead of Tailpipes

I. Introduction & Episode Roadmap

On April 1, 2026, a company that had carried the same name since the year after the First World War quietly stopped being called what it was called. NGK Insulators, Ltd. — 日本碍子 Nippon Gaishi Kaisha, literally "Japan Insulator Company" — became NGK Corporation.¹

Corporate name changes are usually the least interesting thing a company does. This one is worth pausing on, because of the reason management gave for it. Insulators — the white porcelain discs and bells that hang from every high-voltage transmission tower in the world, the product the company was literally founded to make — now account for less than 10% of group sales. Roughly 70% of revenue comes from outside Japan, and more than 60% of employees hold a passport other than a Japanese one.² The name had become a description of the past.

What replaced it is stranger and more interesting than "diversified industrial." NGK today sits at the intersection of two of the largest industrial transitions of this decade, pulling in opposite directions. Its cash engine is the ceramic honeycomb that lives inside the catalytic converter of nearly every internal combustion vehicle on earth — a business whose end market is, on a multi-decade view, dying. Its growth engine is the ceramic hardware inside the machines that make AI chips: electrostatic chucks that hold silicon wafers flat inside etching chambers, aluminum nitride heaters, translucent alumina carriers used to assemble chiplets. One business is a beautifully profitable annuity on a declining asset. The other is a cyclical, capital-hungry ride on the most crowded capex trade in the world.

And in between, in October 2025, management did something that Japanese industrial companies are famous for not doing: it killed a flagship. After roughly four decades of development, deployment and evangelism, NGK's board resolved to stop making and selling NAS sodium-sulfur batteries — the grid-scale storage technology it had dominated globally — and to stop taking new orders entirely.³ The estimated cost was about ¥18 billion.³

The numbers behind all this were good. For the fiscal year ended March 31, 2026, NGK reported net sales of ¥670.1 billion, up 8% year on year, with operating income of ¥95.0 billion, up 17% — records on both lines, achieved despite ¥20 billion of extraordinary restructuring charges.¹⁴ Net income attributable to owners reached ¥59.9 billion.⁴ The market noticed: the shares rose roughly 161% over the twelve months to early August 2026, taking market capitalisation to about ¥1.60 trillion.⁵

So here is the question this piece is built around. Is NGK a genuine transformation story — a 107-year-old materials-science platform successfully rotating from tailpipes to transistors — or is it a well-run legacy manufacturer that happened to own a nice semiconductor components business right when the AI capex wave broke, and is now being priced as though the good part is the whole company?

The story runs in three businesses with three different trajectories, one insider CEO trying to thread them together, one governance blemish that deserves more attention than it usually gets, and a valuation that has already moved a long way. It starts, as most Japanese industrial stories do, with somebody else's factory.

↑ Back to Top

II. Origins: From "Japan Insulator Company" to Ceramics Platform

Nagoya in 1919 was a pottery town. It had clay, kilns, and a craft tradition — and it had 日本陶器 Nippon Toki, the porcelain maker that the world would later know as Noritake, turning out dinnerware for export. Inside Nippon Toki was an odd little division making something that was not tableware at all: electrical porcelain insulators, the ceramic hardware that lets a high-voltage line hang from a steel tower without electrocuting the tower.

Japan was electrifying. Long-distance transmission demanded insulators that could hold tens of thousands of volts, survive typhoons and salt air, and not crack. That was a materials problem, not a dishes problem, and in 1919 the division was spun out as its own company.⁶ The name was pure function: Japan Insulator Company. Nobody in 1919 was building a brand.

The pattern that defines NGK shows up almost immediately, and it is worth naming early because everything afterwards is a variation on it. NGK's actual competency was never insulators. It was the ability to formulate a ceramic body, shape it precisely, and fire it in a kiln so that it comes out of a 1,400°C oven with the properties you designed and the dimensions you specified. Ceramics are unforgiving in a way metals are not: they shrink as they fire, they crack if the thermal gradient is wrong, and you cannot machine your way out of a mistake afterwards. Process knowledge — how to mix, extrude, dry and fire — is the product.

The first redeployment came fast. By 1930 NGK was making spark plugs, another high-temperature, high-precision ceramic component, and in 1936 that business was transferred to a separate company entirely: NGK Spark Plug, 日本特殊陶業 Nippon Tokushu Togyo, renamed ニテラ Niterra in 2023.⁶

This deserves a flag, because it causes genuine confusion to this day. NGK Insulators (now NGK Corporation, ticker 5333) and NGK Spark Plug (now Niterra, a separately listed company) share initials, share a Nagoya heritage, and both sell ceramic components into the automotive supply chain. They are not the same company and have not been for ninety years. Niterra's subsidiary NTK Ceratec even competes with NGK in semiconductor ceramics — two descendants of the same 1919 ancestor selling into the same fab equipment makers. When you read industry share tables that list "NGK" and "NTK" as separate lines, that is why.

The post-war decades were the rebuild: Japan's grid expanded, NGK's insulators went with it, and the company became a quiet utility-adjacent supplier with a strong domestic franchise and steadily improving process technology. Nagoya helped. The city and its surrounding Aichi prefecture had become a dense industrial cluster — ceramics, machine tools, and above all automotive, with Toyota an hour's drive away. A materials company embedded in that cluster gets something no amount of R&D spending buys: proximity to demanding customers who will tell you exactly why your part failed.

The interesting decision was to treat the kiln as the asset rather than the insulator. Through the 1960s and 1970s, NGK's R&D organisation pushed into what the Japanese industry calls ファインセラミックス fine ceramics — engineered ceramic materials for applications with nothing to do with power lines. The distinction matters more than it sounds. Traditional ceramics start with natural clay and tolerate variability. Fine ceramics start with synthesised powders of controlled purity and particle size, and the entire value of the product lies in reproducing the same microstructure batch after batch. Making that transition is a capability change, not a product-line extension: it requires characterisation equipment, statistical process control, and a research organisation that thinks in phase diagrams. NGK made it, and everything the company sells today — substrates, sensors, chucks, wafers — descends from that shift rather than from the insulator itself.

That reframing is the through-line of the whole story. NGK has never really been an insulator company, a catalyst-substrate company, a battery company, or a semiconductor components company. It has been a ceramics R&D and manufacturing platform that has, roughly once a generation, found a new end market willing to pay a premium for materials that survive heat, voltage and precision tolerances at once. The insulator was version one. The next version arrived in the 1970s and turned out to be one of the more consequential industrial inventions of the twentieth century — and one that, fifty years later, still pays most of the bills.

↑ Back to Top

III. Two Legacy Bets That Explain the Culture: HONEYCERAM and NAS

Picture the problem an engineer faced in the early 1970s. The US Clean Air Act amendments had made catalytic converters effectively mandatory on American cars. A catalytic converter works by passing hot exhaust gas over a precious-metal catalyst; the more catalyst surface the gas touches, the more carbon monoxide and hydrocarbons get converted. The early approach — a bed of ceramic pellets — worked, sort of, and then rattled itself to death, choked airflow, and cost the engine power.

The elegant answer was a single ceramic block extruded with thousands of parallel square channels running through it, like a honeycomb, coated in catalyst. Enormous surface area, almost no restriction to flow, one solid piece. The manufacturing problem was brutal. The block has to be extruded with walls measured in fractions of a millimetre, dried without warping, and fired without cracking — and then it has to sit inches from an engine, cycling between ambient temperature and 900°C thousands of times, for the life of the vehicle, without shattering. The material that makes this possible is cordierite, whose near-zero thermal expansion means it barely changes size when heated. Cordierite honeycomb is, in a real sense, a ceramic that has been engineered to ignore temperature.

NGK registered HONEYCERAM as a trademark in 1976 and put ceramic substrates for automotive catalytic converters into production.⁶ Half a century later, cumulative HONEYCERAM output passed two billion units.⁶ Two billion is not a marketing number; it is the physical evidence of a manufacturing learning curve nobody has been able to buy their way onto.

The second bet came out of the same instinct — find a hard materials problem nobody else can solve — and ended very differently.

In the 1980s Japan's utilities faced a structural nuisance: electricity demand collapses at night and peaks in the afternoon, but generating capacity has to be sized for the peak. Store energy overnight and you avoid building plants you use for a few hundred hours a year. NGK pursued joint research with 東京電力 Tokyo Electric Power Company (TEPCO) on a sodium-sulfur battery — a chemistry that runs hot, around 300°C, with molten sodium on one side, molten sulfur on the other, and a fine ceramic solid electrolyte in between doing the work of separating them while letting sodium ions through.⁷

That ceramic separator was the whole game, and it was exactly NGK's kind of problem: a thin ceramic tube, operating in molten sodium at 300°C, for fifteen years, without failing. A prototype NAS system connected to the grid began operating in 1992; mass production started in 2003.⁶ NGK became the dominant global supplier of megawatt-scale, multi-hour storage, shipping container systems rated at 800kW and 4,800kWh and package systems at 1,200kW and 8,640kWh.⁷ By the time of the exit, more than 5 GWh of NAS had been deployed worldwide — the second-most-deployed battery storage technology after lithium-ion, and after pumped hydro the most-deployed medium-to-long-duration storage on the planet.⁸ For two decades, NAS was the "hidden growth story" that analysts told each other about NGK.

Then, at around 7:20 a.m. on September 21, 2011, a NAS installation caught fire. The batteries were owned by TEPCO and installed at the Tsukuba Plant of 三菱マテリアル Mitsubishi Materials in Joso City, Ibaraki Prefecture. Fire authorities declared it under control that afternoon. NGK's investigation traced the cause to a manufacturing defect: within a module of 384 cells, a single faulty cell breached and leaked molten material, which flowed over the sand filler between blocks and shorted cells in an adjoining block.⁹

What NGK did next is the part worth remembering. It halted operations, published a technical account of the cause, recalled the installed base to retrofit safety modifications, redesigned the module architecture, and resumed production the following year.⁹ One defective cell in a 384-cell module became a global product recall — an expensive, credibility-preserving response rather than a cheap one.

The two bets belong together because they came from the same culture: a long-horizon R&D organisation that goes looking for problems where the ceramic is the bottleneck, accepts a decade of development, and then defends the position with process know-how. That culture produced a fifty-year annuity in HONEYCERAM and a technology leadership position in NAS.

It also produced two businesses whose fate was decided by forces entirely outside NGK's control. HONEYCERAM's addressable market shrinks every time an EV replaces a combustion car. NAS's addressable market was taken not by a better sodium-sulfur battery but by lithium-ion's cost curve, which fell faster and further than anyone at NGK modelled in 2005. Great materials science is not the same thing as a durable market position. Hold that thought — it is the spine of the investment case.

Myth vs reality: did the 2011 fire kill NAS?

The convenient story is that the Tsukuba fire destroyed the NAS franchise, and that everything after 2011 was decline. The record does not support it. NAS came back: the recall was completed, the redesign shipped, production resumed, and NGK went on to accumulate the multi-gigawatt-hour installed base that made it the second-most-deployed storage chemistry in the world.⁸ A technology that had been fatally discredited by a safety event does not get deployed at that scale for another decade.

What actually killed NAS was economics, and specifically a scale problem. Sodium-sulfur has genuine engineering advantages for long-duration storage — cheap and abundant raw materials, no lithium supply chain exposure, high energy density relative to lead-acid, a footprint roughly a third the size, long cycle life.⁷ But it requires a hot, insulated, precisely engineered system built around a ceramic component that only one company in the world could reliably make, and its unit cost falls only with volume. Lithium-ion, meanwhile, was being scaled by the consumer electronics and then the automotive industries — end markets thousands of times larger than grid storage — and dragged its costs down on somebody else's volume. NGK was trying to win a cost race against an industry that was being subsidised by the entire global EV build-out.

The lesson generalises well beyond batteries: when your technology's cost curve depends on your own volume and your competitor's depends on an adjacent industry's volume, you are not in a fair fight, and no amount of engineering superiority fixes it. That is the real reason NAS ended, and it took NGK roughly a decade longer than the arithmetic did to accept it.

Before we get to how those two stories resolved, there is a third piece of history that most write-ups on NGK skip, and it says something uncomfortable about how the profitable business was run.

↑ Back to Top

IV. A Governance Reckoning: The Global Auto-Parts Cartel Case

In September 2015, NGK Insulators agreed to plead guilty in a United States federal court and pay a criminal fine of $65.3 million.¹⁰¹¹ The charge was conspiracy to fix prices and rig bids for ceramic substrates for automotive catalytic converters — the HONEYCERAM franchise itself — sold to automakers including General Motors, トヨタ自動車 Toyota Motor and 日産自動車 Nissan Motor, and to their subsidiaries and suppliers, in the United States and elsewhere. The Justice Department dated the conspiracy from at least July 2000 until at least February 2010.¹⁰

That alone would place NGK inside the largest criminal antitrust investigation in US history: by that point 36 companies and 30 executives had pleaded guilty or agreed to plead guilty in the auto-parts probe, with more than $2.5 billion in criminal fines.¹⁰ Being one of 36 is, in a grim way, mitigating context.

The second count is not. NGK also pleaded guilty to obstruction of justice for altering, destroying, mutilating and concealing documents between February 2010 and approximately July 2012, with the intent of impeding the investigation.¹⁰ Read those dates carefully. February 2010 is roughly when the price-fixing stopped. The document destruction ran for two and a half years after — that is, while the company knew it was under investigation.

Why does a 2015 case belong in a 2026 investment discussion? Three reasons, and none of them is moralising.

First, it locates where the cartel behaviour happened. This was not a rogue overseas distributor or an obscure subsidiary. It was the core profit engine — the substrate business that then, as now, funded everything else. When a company's single most profitable franchise is also the one implicated in a decade-long price-fixing conspiracy, an investor is entitled to ask a mechanical question: how much of that historically fat margin was engineering and how much was coordination? Nobody can answer that precisely. But it is the right question to hold in mind when Section V describes a "technically defended" duopoly-plus with structurally high margins.

Second, obstruction is a culture signal in a way that price-fixing alone is not. Cartels can persist in a business unit for years without the board knowing. Systematically destroying documents during an active investigation implies people believed concealment was a viable institutional strategy. That is a statement about internal control and tone, and it is precisely the sort of thing governance reform is supposed to fix.

Third — and this is where the analysis has to be honest about its own limits — NGK did respond. Following the plea, the company disclosed the agreement and, per its own account, strengthened competition-law compliance: a business ethics committee weighted toward outside directors, a group compliance function, an externally managed whistleblower hotline, and an independent committee on competition-law compliance drawing on external directors, external auditors and outside counsel.¹¹ Those are the right boxes. They are also exactly the boxes every company ticks after a cartel case, which makes them weak evidence on their own.

The stronger evidence for a sceptic is behavioural, and it cuts in NGK's favour: eleven years on, there has been no publicly disclosed repeat antitrust action against NGK of comparable scale, and the company's more recent disclosures are notably franker about unprofitable businesses than the pre-2015 vintage. The weaker evidence is presentational. NGK's current investor and sustainability communications largely present compliance in forward-looking, systems-language terms rather than confronting the 2015 case head-on. That is normal corporate practice, and it is also a small missed opportunity — companies that name their own worst moment tend to earn more credibility on governance than companies that describe their committee structure.

It is also worth understanding why the automotive components industry produced such an extraordinary volume of cartels — 36 companies is not a coincidence, it is a pattern. The market design almost invites it. Components are sourced through multi-year request-for-quote processes for specific vehicle platforms; the same handful of suppliers bid repeatedly against each other for decades; volumes and timing are announced years ahead; and the buyers are large enough to squeeze margins relentlessly. Suppliers in that structure end up in frequent, legitimate contact — at industry bodies, at joint development programmes, at customer quality reviews — while facing a shared incentive to stop competing. It takes an active compliance culture to keep that from curdling. NGK's did not, for a decade.

For an investor, the practical takeaway is narrow and durable: in a concentrated supplier industry, with a handful of global players selling qualified components to a handful of global buyers, antitrust risk is structural, not historical. It is a feature of the market NGK operates in — and, notably, of the semiconductor components market it is now expanding into, which shares the same profile of few sellers, few buyers, long qualification cycles and high price visibility. Which is a useful way to walk into that market.

↑ Back to Top

V. The Core Engine Today: Environment Business (Automotive Ceramics)

Every argument about NGK's future has to start by respecting its present. The Environment business — automotive ceramics plus a smaller industrial-process line — generated ¥399.5 billion of sales in the year to March 2026 and ¥68.6 billion of operating income, a 17.2% operating margin.¹ Against group totals of ¥670.1 billion and ¥95.0 billion, that is roughly 60% of revenue and, before corporate costs, the overwhelming majority of profit.¹ Everything else NGK does — the AI capex story, the new plants, the buybacks, the R&D budget — is funded out of the exhaust systems of internal combustion vehicles.

What is actually in the box

The product line is best understood as the full exhaust-purification stack. HONEYCERAM substrates are the catalyst carriers already described. Diesel particulate filters — including silicon-carbide DPFs for heavy-duty applications — are honeycombs with alternate channels plugged, forcing exhaust gas through the porous ceramic wall so soot is physically trapped. Gasoline particulate filters (GPF) do the same job for direct-injection petrol engines, a category regulators only recently decided to care about. And NOx and gasoline sensors are ceramic electrochemical devices sitting in the exhaust stream, telling the engine control unit in real time what the gas composition is so the system can adjust.

The FY2026 plan gives the shape: within roughly ¥400 billion of Environment sales, management guides to about ¥92 billion of honeycomb substrates, ¥76 billion of sensors, ¥74 billion of silicon-carbide DPF, ¥76 billion of truck and off-road filtration, ¥57 billion of GPF, and ¥25 billion of industrial process products.¹ Note what that mix says: the classic catalyst substrate is now less than a quarter of the segment. NGK has spent two decades adding higher-value content around it.

Industry structure: why this is not a commodity

The market is genuinely concentrated. Third-party market research puts Corning Incorporated in the lead with roughly 25–30% of ceramic honeycomb substrate revenue, with NGK's ceramics operation and Japanese peer イビデン Ibiden together accounting for close to 40%.¹² Add デンソー Denso and Saint-Gobain and the top five hold most of the global market.¹² Treat any single share estimate with appropriate scepticism — these are analyst constructions, not audited disclosures — but the directional picture is corroborated by the customer side: automakers do not have many places to go.

The moats here are worth stating precisely, because "high switching costs" gets used lazily. A substrate is not a bolt-on part; it is qualified into a specific vehicle platform's emissions certification, years before the vehicle launches, alongside the catalyst formulation and the engine calibration. Changing supplier mid-programme means re-certifying emissions compliance in every regulated market where the platform sells. Automakers will do that for a big enough saving, but the bar is high and the lead time is long. Layer on top: fifty years of yield learning on a process where scrap is the main cost variable, and installed extrusion and kiln capacity sited near customers on three continents.

The honest counterweight is that none of this is permanent. Chinese producers — names like Jiangxi Kexing and Pingxiang Chemshun — are expanding, and they are attacking exactly where an incumbent is most vulnerable: a mature product, in a large domestic market, where the local regulator sets the specification.¹² China VI compliance is a Chinese specification problem, and Chinese suppliers get to learn on the largest vehicle fleet in the world.

The tension management cannot resolve

Here is the thing NGK's own materials say out loud, which is more than many companies in its position manage. In the FY2026 outlook, management stated that Environment shipments will decline because of the gradual shift to electric vehicles, while sales are expected to rise as the mix moves to higher value-added products — and that segment income would be roughly flat, at about ¥67 billion on ¥400 billion of sales, a margin of 16.8% versus 17.1% the prior year.¹

Unpack that. Units down. Revenue up. Profit flat. That is the arithmetic of a business running up an escalator: content per vehicle rising fast enough to offset falling vehicle counts, but not fast enough to also grow profit, especially with raw material and energy costs rising — management explicitly baked ¥2.0 billion of extra energy cost into FY2026 guidance because of crude oil price pressure tied to the Middle East situation.¹

The mix-shift tailwind is real and regulator-driven. Tightening standards — Euro 7 in Europe, Tier 4 with a 0.5mg particulate limit in the US, China 7, BS7 in India — mean more sensors, more GPF, more filtration content per vehicle, and management expects tightened European and US rules to drive sensor demand from FY2027 with GPF ramping thereafter.¹ More stringent rules make each remaining combustion vehicle worth more to NGK.

The offsetting fact is that the escalator only runs so long. Interestingly, NGK's own vehicle forecast moved in its favour over the past year: management cut its estimate of non-combustion (EV and fuel cell) vehicles as a share of 2030 global sales from 34% to 30%.¹ That is a small piece of evidence that the EV transition is running slower than the 2021–2023 consensus assumed — a genuine near-term reprieve, and one that hybrids extend further, since a hybrid still has an exhaust system and often a more demanding one, because the catalyst has to stay hot through engine stop-start cycles.

But a reprieve is not a reversal. On any reasonable path, combustion-linked ceramic volumes decline over the next two decades. The question is not whether but how fast, and whether NGK can raise content per vehicle faster than units fall for long enough to matter.

Myth vs reality: is the Environment business already rolling over?

The consensus shorthand on NGK for most of the past decade was simple: EVs are coming, catalytic converters are going, and the automotive ceramics business is a melting ice cube. It is worth checking that against what actually happened.

Environment sales rose from ¥390.4 billion in FY2024 to ¥399.5 billion in FY2025, with operating income of ¥68.3 billion and ¥68.6 billion respectively, at margins of 17.5% and 17.2%.¹ Guidance for FY2026 is ¥400.0 billion of sales and ¥67.0 billion of income.¹ Three years of flat-to-slightly-up revenue and flat profit at a ~17% margin is not an ice cube melting. It is a plateau.

Two mechanisms explain the gap between narrative and results. Hybrids have grown faster than pure EVs in most markets, and a hybrid needs the same exhaust after-treatment content as a combustion car — arguably more, because a catalyst that keeps cooling down when the engine shuts off is a harder thermal problem than one that runs hot continuously. And regulation keeps adding content: gasoline particulate filters did not exist as a required product a decade ago and are guided to ¥57.0 billion of FY2026 sales on their own.¹

The reality is therefore neither the bull nor the bear version. Environment is not growing, and management does not claim it is. It is also not collapsing, and it is throwing off enough cash to fund a ¥450 billion five-year investment programme aimed elsewhere.¹ The correct frame is a high-margin plateau of uncertain length — and the investor's job is to form a view on how long the plateau lasts, not on whether the decline has started.

The adjacency bet

Management's structural answer is to reuse the same materials science somewhere else. Two candidates recur: sub-nano ceramic membranes for gas separation — the same porous-ceramic competency, applied to separating nitrogen and CO₂ — and honeycomb structured sorbent materials for direct air capture, which is literally the HONEYCERAM geometry repurposed to pull CO₂ out of ambient air.¹

Evaluate these honestly. They are technically credible, because the manufacturing asset genuinely transfers. They are also, today, not material. NGK groups them under a carbon-neutrality bucket targeting ¥10 billion or more of sales by FY2030 within its new-business programme — meaningful as optionality, immaterial against a ¥400 billion segment.¹ Direct air capture in particular depends on a carbon price and a policy regime that does not yet exist at scale. This is a real option, not a plan.

The investor conclusion for the Environment business is uncomfortable but clear. This is a high-quality, structurally advantaged, cash-generative franchise attached to a shrinking end market, and the company is being paid to harvest it well rather than to grow it. Management is doing what you would want — pushing mix, rationalising production, holding margin near 17% — but no amount of mix shift converts a declining market into a growing one. Which is precisely why the second segment matters so much more than its size suggests.

↑ Back to Top

VI. The Growth Engine: Digital Society Business — Riding the AI Capex Wave

Inside a plasma etch chamber, a silicon wafer has to be held perfectly flat, at a precisely controlled temperature, in a vacuum, while an aggressive corrosive plasma removes material from its surface a few atoms at a time. You cannot clamp it mechanically — clamps distort the wafer and shed particles. So the wafer is held by electrostatic force, using a ceramic disc with electrodes embedded inside it that generates the attraction. That disc is an electrostatic chuck, and it has to be flat to within microns, thermally uniform, electrically exact, and chemically inert against the plasma. If it fails, or drifts, every wafer on the tool is compromised.

That is NGK's second act, and it is exactly the same problem as the insulator and the catalyst substrate, restated: build a ceramic that does something impossible under extreme conditions, repeatedly, at scale.

The numbers, and what changed

The Digital Society segment produced ¥205.4 billion of sales in the year to March 2026, up from ¥171.6 billion, with operating income of ¥28.1 billion against ¥17.2 billion — revenue up roughly 20%, profit up roughly 63%, and operating margin expanding from 10.0% to 13.7%.¹ Management guides FY2026 to ¥245.0 billion of sales and ¥36.0 billion of operating income, a 14.7% margin.¹

The margin expansion is the analytically important part, more than the growth. Revenue growth in a capex upcycle tells you the customer is buying. Margin expanding by nearly four points while revenue grows 20% tells you something about operating leverage and, plausibly, about pricing — this is a business where fixed costs are heavy (kilns, cleanrooms, machining) and incremental volume drops through hard. It is consistent with a supplier that is capacity-constrained rather than price-constrained. But it is not proof of pricing power, because you cannot separate cycle from structure inside a single upcycle. The test comes on the downswing.

Within the segment, the semiconductor production equipment (SPE) components line — chucks, aluminum nitride heaters, and related parts — was ¥143.3 billion of the FY2025 total, growing to a planned ¥165.0 billion.¹ Electronic components, at ¥33.9 billion, are guided to ¥50.0 billion, the fastest-growing line, and metal-related products round out the rest.¹

What is actually driving it

Three distinct demand vectors, which is more diversification than "AI" suggests.

The first is front-end wafer processing. As chips get denser and more three-dimensional — more layers, tighter features — the number of etch and deposition steps per wafer rises, which raises the number of chambers, which raises demand for chucks and heaters. NGK's own materials frame the SPE component opportunity against a semiconductor front-end equipment investment cycle rising through FY2027, with the latest industry forecast revised upward versus October 2025.¹

The second is advanced packaging, and it is the more interesting one. When a chipmaker builds a large AI processor out of multiple smaller dies — chiplets — those dies have to be assembled on a temporary carrier during processing. Glass has been the conventional carrier, and glass warps and breaks. NGK's translucent alumina wafer is stiffer and more durable, cutting warpage and breakage in manufacture.¹³ In November 2025 the company announced it would triple production capacity by fiscal 2027 across its Komaki site in Aichi Prefecture and Mine site in Yamaguchi Prefecture, targeting ¥20 billion of sales into the high-performance semiconductor market by FY2030.¹³ Separately, the HICERAM Carrier product — used in chiplet assembly — is being expanded to roughly three times current capacity by FY2027, with management aiming to exceed ¥10 billion of sales in FY2026 and reach ¥25 billion by 2030.¹

The third is mundane and useful: piezoelectric micro-actuators for hard disk drives, the tiny ceramic elements that position a read head with nanometre precision. Data centres still buy an enormous volume of spinning disks for capacity storage, and management describes this demand as stable.¹

The capex commitment — this is where talk becomes evidence

Between FY2021 and FY2025, NGK invested ¥82.7 billion of capital in the Digital Society business. The plan for FY2026 through FY2030 is ¥250.0 billion — roughly triple.¹ The centrepiece is a new NGK Ceramic Device plant on land adjacent to the existing Ishikawa Plant: approximately ¥70 billion of investment, construction starting April 2027, operations from October 2029, lifting group capacity for semiconductor manufacturing equipment components by about 20%.¹ Segment sales are targeted at roughly ¥350 billion by FY2030 against ¥205.4 billion today.¹

This is the single most consequential capital allocation decision in the story, and it deserves to be read both ways. On the bull side: ¥250 billion of committed capex is not a slide, it is a decision with concrete plant sites and dates, and management is putting the money where the returns are — a segment whose internal ROIC sits above the newly raised 12% hurdle rate.¹ On the bear side: a plant that starts operating in October 2029 was sanctioned against a demand forecast made in 2026, in the most forecast-abused industry in the world. Semiconductor equipment capacity built at the top of a cycle has a long history of arriving into a trough. The mitigation is that this capacity is component capacity, not fab capacity — smaller cheques, shorter lead times, more increments — but the risk is real and management should be judged on whether it phases the spend or ploughs ahead when orders wobble.

The competitive question

NGK is a leader here, not the leader. Independent market work on electrostatic chucks names 新光電気工業 Shinko Electric Industries, NGK, TOTO, 京セラ Kyocera and NTK Ceratec as the top five, collectively holding more than a third of global revenue, with Entegris, Sumitomo Osaka Cement and expanding Chinese specialists such as Beijing U-Precision behind them.¹⁴ Note the irony flagged earlier: NTK Ceratec belongs to Niterra, the 1936 spin-off.

Two implications follow. First, the moat here is narrower than in automotive substrates. There is no emissions-certification lock-in; qualification with a tool maker is sticky but not regulatory. Second — and this is the sharper point — NGK's customers are Applied Materials, Lam Research, Tokyo Electron and their peers: a small number of very large, very sophisticated buyers who dual-source deliberately and who have the engineering capability to develop chucks internally. That is a structurally strong buyer set. It caps how much of the AI boom's economics a component supplier can retain.

Myth vs reality: is NGK an "AI stock"?

The share price behaved like one over the past year, and the segment growth justifies part of that. But the framing deserves a fact-check on three counts.

First, proportion. Even after 20% growth, Digital Society was about 31% of group revenue in FY2025 and is guided to roughly 35% in FY2026.¹ The majority of NGK's revenue remains attached to combustion engines. An investor buying NGK for AI exposure is buying roughly one-third pure-play and two-thirds something else — with the something else being the part that funds the AI capex.

Second, position in the value chain. NGK does not sell to AI companies. It sells consumable and semi-consumable components to the makers of the machines that make the chips. That is two steps removed from the end demand, which means the demand signal arrives late and leaves late — orders reflect tool builds, which reflect fab decisions, which reflect chip demand forecasts. Being late in the chain is protective on the way into a downturn and punishing on the way out of one.

Third, what "AI-driven" is doing inside the numbers. Management attributes the FY2025 result partly to semiconductor demand for AI applications and partly to a concentrated inventory build by certain customers.¹ That second clause is important and easy to skim past. Inventory replenishment inflates a supplier's revenue relative to true end consumption, and it reverses. Any component supplier reporting outstanding growth during a customer restocking phase should be assumed to have borrowed some demand from the following year.

None of this makes the growth fake. It makes the growth derived — a leveraged, lagged, partially inventory-inflated read on semiconductor capital spending, dressed in AI language because AI is the reason the spending is happening. That is a genuinely attractive place to sit in an upcycle. It is a specific and knowable set of risks, not a general one.

EnerCera, briefly

One aside deserves a mention and no more. EnerCera is NGK's chip-scale ceramic rechargeable battery for IoT devices and smart cards — a genuinely novel product built on the same ceramic electrode competency, sampled by hundreds of companies. It is not financially material and does not appear as a named growth pillar in the FY2030 plan. File it as real technology, unproven commercial scale.

The investor conclusion here is that Digital Society is the reason NGK re-rated, and it deserves most of that credit — the growth is visible in reported profit rather than promised in a roadmap. But it is a cyclical business with a concentrated, powerful customer base, and the company is committing triple the historical capital into it at a point in the cycle when everyone else is doing the same. That combination has been kind to shareholders for a year and has historically been unkind eventually.

Which brings us to the segment where NGK has just demonstrated what it does when a long bet stops working.

↑ Back to Top

VII. The Third Segment and the NAS Exit: Energy & Industry Business

On October 31, 2025, NGK's board met and resolved to discontinue the manufacturing and sales of NAS batteries under its Energy Storage Business, and to stop accepting new orders.³ The estimated cost, roughly ¥18 billion, was booked as an extraordinary loss.³¹⁵ In the event, total business-restructuring charges for the fiscal year came to ¥20 billion, partially offset by ¥13.2 billion of gains on sales of investment securities — a classic Japanese balance-sheet manoeuvre, funding a restructuring by unwinding cross-shareholdings.¹

The proximate cause was a partner walking away. NGK had been in discussions with BASF — its partner since 2019 — on expanding supply capacity and reducing costs in anticipation of energy storage demand growth. Those discussions ended in September 2025.³ Industry reporting attributed BASF's withdrawal to its refocus on core segments under a new chief executive, and described an unsuccessful search for alternative financing in which a prospective investor pulled back, with the collapse of lithium-ion venture Northvolt cited as having chilled appetite for upstream battery technology bets.⁸

The scale of what was closed

NAS generated about ¥6.4 billion of sales in the fiscal year — roughly 1% of group revenue.⁸ That framing is important and slightly deflating: this was not a large business by the end. It was, however, a large idea, with more than 5 GWh deployed globally, and it was the reason the Energy & Industry segment existed in its old shape.

The segment is the smallest of the three. On the reclassified basis introduced for FY2026 — radioactive waste treatment systems moved over from Environment — Energy & Industry recorded ¥73.2 billion of FY2025 sales, comprising roughly ¥51.0 billion of insulators, ¥11.8 billion of energy storage and ¥10.5 billion of energy plant, with operating income of just ¥0.3 billion, a 0.3% margin.¹ On the old basis, it lost ¥1.3 billion.¹ For FY2026, management guides to ¥65.0 billion of sales and ¥4.0 billion of operating income — a 6.2% margin, achieved essentially by removing the loss-maker.¹

That is the cleanest possible demonstration of what the NAS business had become: subtracting it turned a segment that lost money into one that makes a passable if unexciting return.

Read it both ways

The capital discipline reading. NAS was capital-intensive, sub-scale, and out-competed on cost by a lithium-ion industry whose learning curve NGK could not match. Sodium-sulfur's advantages — long duration, no lithium supply exposure, long cycle life — were real, but lithium-ion's cost decline swamped them, and the technology's obvious remedy was exactly the one BASF was going to fund: scale. When scale funding vanished, continuing would have meant NGK financing a gigawatt-scale battery ramp alone, on its own balance sheet, against entrenched competition. Management raised its internal ROIC hurdle from 10% to 12%, introduced a formal business restructuring and divestiture review process screening on operating income, ROIC and free cash flow, and then applied it to its own prestige product.¹ That is not nothing. Companies that build such processes and never use them are common.

The capital allocation failure reading. Lithium-ion's cost trajectory was not a surprise in 2025. It was visible from roughly 2015 onward. NGK spent the intervening decade continuing to invest in a technology whose economic case was eroding annually, and only acted when an external partner forced the issue. The company's own disclosure is candid on the consequence: ROE fell to 6.1% in FY2022 and stagnated at 7.8% through FY2024 and FY2025, and management attributes that stagnation in part to delays in exiting and restructuring unprofitable businesses.¹ That is a striking admission to put in a results deck — and it is the strongest evidence in the file that this exit was late rather than decisive.

The fairest synthesis: the decision itself was correct and the process behind it looks genuine, but the timing was reactive. NGK did not conclude that NAS had lost; it concluded that NAS had lost once its partner stopped paying. An investor assessing management credibility should weight the speed of the second decision more than the merits of the first — and here there is a real datum. In the same fiscal year, NGK also decided to close the Chita Plant in the insulator business, accepting lower near-term utilisation and a year-on-year profit decline in insulators to do it.¹ Two restructurings in one year, both self-inflicted pain, is a pattern rather than a one-off.

What remains

The rest of the segment is the 1919 business plus engineering. Insulator demand is, unexpectedly, decent: management points to steady transmission and distribution network investment in Japan and overseas, driven partly by data centre buildout.¹ The AI boom is, indirectly, propping up NGK's oldest product line — grid capacity has become the binding constraint on data centre expansion, and grid capacity requires insulators.

Alongside it, NGK established a new Energy Plant Division from FY2026, consolidating the low-level radioactive waste treatment business transferred from Environment with hot-line insulator washing equipment, aiming to build engineering and construction capability in power-related fields.¹ That is a sensible tidy-up. It is also a reminder that this segment is now a collection of stable, unglamorous, low-growth businesses rather than a growth story. After the exit, Energy & Industry's job is to earn its cost of capital and stop being a distraction.

What to press management on

The exit disclosure and the results materials answer the what clearly and the why now only partially. Three questions remain genuinely open, and they are the ones worth listening for on future calls.

How much residual obligation remains? NGK has committed to continuing service for existing installations, which for a fifteen-year-life product installed as recently as 2025 implies a service tail running well into the 2030s.⁸ The ¥18 billion estimate presumably provides for it, but the provision's adequacy is an accounting judgment, and provisions for long-tail service obligations have a habit of being topped up.

What happens to the assets and the people? Sodium-sulfur manufacturing plant is highly specialised; the ceramic electrolyte tube line in particular has no obvious alternative use. Redeploying the workforce into a growing Digital Society business is the logical answer and the one management has signalled, but redeployment across different plants, prefectures and process disciplines is not costless.

And what does the decision imply for the remaining new-business pipeline? NGK's growth programme still contains ceramic membranes, direct air capture sorbents and an energy solutions business — all of them long-horizon, capital-hungry bets on markets that do not yet exist at scale, which is precisely the profile NAS had in 1990.¹ The company has just demonstrated that it will eventually stop funding such a bet. The more valuable demonstration would be stopping one earlier.

The larger question the NAS decision raises is about the person who made it.

↑ Back to Top

VIII. Management, Capital Allocation, and the Name Change

Shigeru Kobayashi joined NGK in 1983 and became president in April 2021 — a 38-year internal ascent through the insulator and power businesses before reaching the top job.¹⁶⁶ This matters for how you read the strategy. Kobayashi is not an outside disruptor brought in to break the company; he is a career insider who spent his formative decades in the segment that has since become the smallest and least dynamic of the three, and who is now presiding over the reallocation of capital away from it.

That background cuts both ways. Insiders know where the bodies are buried and can move an engineering organisation that would resist an outsider. Insiders also carry the sunk-cost history personally. In Kobayashi's own framing, the constant is the technology base: "We will never waver in this basic approach, even in times of great change. We will not fear challenges when tackling difficult themes, and we will not stop in our technological development."² That is a statement about ceramics, not about any particular end market — which is, at least, internally consistent with a strategy of abandoning batteries while tripling down on semiconductor components.

The record, judged on behaviour

The most useful way to assess a management team is against its own prior promises. NGK published a Group Vision with FY2025 financial targets, and the FY2025 results deck scores itself against them explicitly — which is more accountability than most companies volunteer.¹

The scorecard is mixed in an informative way. Sales came in at ¥670.1 billion against a ¥600.0 billion target; operating income ¥95.0 billion against ¥90.0 billion; EPS ¥206 against ¥200.¹ All beats. But net income was ¥59.9 billion against a ¥60.0 billion target — a miss by a whisker — and ROE was 7.8% against a 10% target, a substantial miss.¹ Management's own explanation is that the sales and profit beats were flattered by a weaker yen (the plan assumed ¥100 to the dollar; actual was ¥151), while ROE fell short because of cost inflation and, again, delayed exits from unprofitable businesses.¹

That is the right way to read a beat. NGK hit its revenue and profit targets substantially on currency, and missed the one target — return on equity — that currency does not rescue. Investors should mark management on the ROE line.

The forward targets are correspondingly more demanding: FY2030 sales of ¥900 billion, operating income of ¥150 billion at a 16.7% margin, net income of ¥100 billion, EPS of ¥370, and ROE of 12%.¹ Getting from 7.8% to 12% ROE requires roughly two-thirds more profit on an equity base that share buybacks will only partly restrain. It is achievable if Digital Society delivers ¥350 billion of sales at improving margin and Environment holds flat; it is not achievable if the semiconductor cycle turns and the substrate business rolls over together.

Narrative consistency across the cycle

One useful discipline when assessing a management team is to check whether the current story is the same story told three years ago with better numbers, or a different story retrofitted to the numbers that happened to arrive.

On this test NGK does reasonably well. The Group Vision published under Kobayashi framed the company as one contributing to carbon neutrality and a digital society through unique ceramic technologies — the two-pillar framing that the current segment structure, the capex allocation and the name change all express.²¹² The strategic language has not lurched. What has changed is the weighting: carbon neutrality was the louder half of the pair in 2021–2022 communications, and the digital half has become the louder one as semiconductor demand delivered and hydrogen, ammonia and direct air capture stayed in development. Management has not disowned the carbon-neutrality leg — NGK reported CO₂ emissions down from 730 thousand tonnes in FY2013 to about 500 thousand tonnes in FY2025, a 32% reduction against a 50% target for 2030 and net zero by 2050, and it announced hydrogen-combustion mass production technology for its own kilns in March 2026 and a joint research agreement with 中部電力 Chubu Electric Power on ammonia-fired kilns in February 2026.¹ But an honest reading of the segment plan is that carbon neutrality is currently a cost-and-compliance programme plus an R&D option, while digital society is the earnings story.

The one place the narrative has genuinely shifted is on portfolio discipline, and management says so rather than pretending otherwise. Framing NAS, the Chita insulator plant and the ceramic packages business as things to be exited, closed or restructured is a different posture from the 2021-vintage emphasis on nurturing new businesses. Companies that change posture and explain why are easier to underwrite than companies whose targets quietly move.

Incentives — flag the thin evidence

On skin in the game, the evidence is thin and worth stating plainly rather than glossing. Third-party compilations of NGK's Japanese securities filings put Kobayashi's total annual compensation at roughly ¥138 million, split close to evenly between fixed salary and performance-linked bonus, with a direct personal shareholding of around 0.014% of shares outstanding. NGK does not present executive compensation or individual shareholdings in its English-language investor materials, so those figures should be treated as indicative rather than verified here; the primary source is the company's Japanese 有価証券報告書 annual securities report.

Even taking them at face value, the analytical point is modest and unsurprising: this is standard Japanese large-cap practice — meaningful bonus linkage, negligible equity ownership. It means the CEO's personal wealth is tied to annual profit far more than to the long-run share price. For a strategy whose payoff arrives in 2029 and 2030, that is a genuine, if conventional, misalignment worth noting.

Capital allocation: the shift is real, the trigger is external

The change in NGK's capital returns policy over the past three years is the clearest observable change in behaviour.

The dividend policy was revised upward to target a dividend-on-equity ratio of 3.5% and a payout ratio of 35% or higher, against previous targets of 3% and 30%.¹ The FY2025 annual dividend was ¥80 per share, a 38.8% payout and 3.0% DOE; the FY2026 plan raises it by ¥26 to ¥106, taking DOE to 3.7%.¹ A 33% dividend increase in one year, from a Japanese industrial, is a statement.

Buybacks have been running annually and escalating: ¥9.6 billion in FY2022, ¥14.9 billion in FY2023, ¥9.4 billion in FY2024, ¥15.0 billion in FY2025, and then — announced alongside the FY2025 results on April 30, 2026 — up to 6.5 million shares for up to ¥33 billion, representing 2.3% of shares outstanding excluding treasury, with all acquired shares to be cancelled on June 1, 2026.¹ Execution was via the off-auction N-NET3 system on the Nagoya Stock Exchange, a nod to NGK's dual listing and Nagoya roots.¹ The cancellation matters: shares repurchased and retired reduce the denominator permanently, unlike treasury shares that can be reissued.

Underneath the returns, NGK has become explicit about capital costs in a way Japanese industrials historically were not. The company now publishes its own estimate of its cost of capital — 9.3% cost of equity on a CAPM basis, 7.9% WACC, and an 11.3% pre-tax WACC used as the internal hurdle — and has raised the business hurdle rate from 10% to 12% ROIC precisely because capital costs rose.¹ Group ROIC on NGK's own definition improved from 9.8% in FY2023 to 13.9% in FY2025, with 15.9% forecast for FY2026.¹ Financial leverage is deliberately managed to a debt-to-equity ratio around 0.3.¹

But who is driving?

Here is the activist's question. NGK has discussed return on equity as a management priority since the 1990s, and its own CFO commentary in 2022 laid out financial and capital strategies to support the transformation.¹⁷ And yet ROE sat at 6.1% in FY2022 and 7.8% in FY2024, well below a ~10% cost of equity — meaning the company was destroying economic value on a book-equity basis for years while talking about capital efficiency.

What changed was not conviction; it was pressure. On March 31, 2023, the Tokyo Stock Exchange formally requested that Prime and Standard Market companies implement management conscious of the cost of capital and share price, aimed squarely at the roughly 43% of large Japanese companies then trading below book value.¹⁸ Disclosure compliance in the Prime Market exceeded 90% by March 2025, and across the market average price-to-book rose from about 1.1 to 1.4 and average ROE from 8.4% to 9.0% over the following three years.¹⁸

NGK's own trajectory maps onto that timeline almost exactly: hurdle rate raised, formal divestiture screening introduced, dividend policy upgraded, buybacks scaled, restructurings executed. As of March 31, 2026, NGK's price-to-book was 1.45 and its P/E 19.83.¹ The company escaped the sub-1x PBR trap — but so did the market average, and at roughly the same time.

The fair conclusion is that NGK's capital discipline is real in its effects and reactive in its origins. That is not damning. External pressure that produces durable behavioural change is still durable behavioural change, and the hurdle-rate increase and divestiture screening process are structural rather than cosmetic. But an investor should not confuse a company that responded well to a regulatory push with a company that would have acted without one. The test comes when the AI cycle turns and shareholder returns compete with capex for the same cash.

The rebrand

Which brings us back to the name. The rationale management gave — insulators under 10% of sales, roughly 70% of revenue overseas, over 60% of employees non-Japanese, and internal and external calls for a change — is entirely reasonable, and the group kept the familiar blue NGK symbol while unifying the Japanese and English names.²¹⁹

But notice what a name change actually is. NGK did not rebrand in 2021 when Kobayashi arrived, or in 2023 when it started raising hurdle rates. It rebranded in 2026, after the Digital Society segment had nearly doubled its profit, after the NAS exit was decided, after the record year was in the bank. The name change is not a promise about the future; it is a ratification of a transformation management judged to be far enough along to sign its name to. Read it as a lagging indicator — which, for an investor, makes it more informative rather than less. Companies that rename themselves before the transformation are making a claim. Companies that rename themselves after are reporting a result.

↑ Back to Top

IX. Bull vs. Bear: The Investment Case

Strip the story down and NGK is two businesses with opposite time-arrows glued to a shared manufacturing culture. The investment question is whether the glue is worth anything — whether being one company rather than two creates value, or merely averages a declining annuity against a cyclical growth asset.

The competitive position, examined properly

Run Porter's framework across both engines and the asymmetry is stark.

Barriers to entry are high in automotive substrates — a combination of regulatory qualification, fifty years of process yield learning, and installed local capacity — and moderate in semiconductor ceramics, where qualification with a tool maker is sticky but the certification lock-in of emissions compliance simply does not exist.

Buyer power is high in both, and this is the most underrated feature of NGK's economics. Global automakers and the handful of semiconductor equipment OEMs are among the most capable procurement organisations in the world; both dual-source by policy; both have the engineering depth to insource if a supplier gets greedy. Whatever else NGK owns, it does not own the customer relationship on its own terms.

Rivalry is disciplined but real: Corning and Ibiden in substrates, Shinko, Kyocera, TOTO and NTK Ceratec in chucks.¹²¹⁴ Substitutes are the existential category — the substitute for a catalytic converter is not a cheaper converter, it is a battery-electric drivetrain that needs none. Supplier power is modest, though raw material and energy costs bite, as the ¥2.0 billion energy provision demonstrates.¹

Through Hamilton Helmer's 7 Powers lens, NGK holds two powers with confidence and gestures at a third. Process power is the strongest and most defensible: two billion HONEYCERAM units of accumulated yield learning is not replicable with capital alone.⁶ Switching costs are genuine in automotive, weaker in semis. Scale economies exist in both but are shared with equally large rivals. What NGK does not have is network economies, branding power in any meaningful commercial sense, or cornered resources. And counter-positioning — the power that matters most against disruption — runs against NGK in automotive, not for it. The EV manufacturer is the counter-positioned entrant; NGK is the incumbent whose asset base is the liability.

The bull case

Digital Society is not a promise; it is showing up in reported profit, with segment operating income up roughly 63% in a single year and margin expanding nearly four points.¹ The capex behind it is committed and specific, with named plants and dates, not a strategy slide.¹ The near-term Environment picture is better than the EV narrative implies, and NGK's own revision of its 2030 non-combustion vehicle forecast from 34% down to 30% is evidence that the runway is longer than the consensus of three years ago.¹ Regulation genuinely raises content per remaining combustion vehicle.¹ Management has demonstrated willingness to take restructuring pain twice in one year, and has translated capital discipline into a materially higher dividend and a scaled buyback with cancellation.¹ And the balance sheet is conservative — operating cash flow of ¥138.0 billion in FY2025, cash of ¥199.6 billion, D/E of 0.3 — which means the ¥450 billion five-year capex plan does not require financial engineering to fund.¹

The bear case

Roughly 60% of revenue and the large majority of profit still comes from a business tied to the internal combustion engine, and management's own FY2026 guidance concedes falling shipment volumes with flat segment profit.¹ Mix shift offsets decline; it does not reverse it, and it eventually runs out of content to add.

Digital Society, the offset, is levered to semiconductor capital expenditure, historically among the most violent cycles in industry. NGK is expanding capacity by roughly 20% into a plant that opens in late 2029, alongside every competitor doing the same.¹ A memory downturn or an AI capex digestion phase would hit the segment that is currently masking Environment's structural problem — and would do so while the depreciation from that expansion is ramping.

Governance carries a real blemish: a $65.3 million criminal antitrust fine plus an obstruction of justice conviction, in the core profit engine, within recent institutional memory.¹⁰ Management incentive alignment is thin on the equity side. The NAS exit, however correct, closed a business built over four decades and was triggered by a partner's withdrawal rather than by NGK's own conviction — and the company itself concedes that delayed exits held ROE back for years.¹

And then there is the price. The shares rose roughly 161% in twelve months, and trade at about 23 times trailing and roughly 18 times forward earnings with a dividend yield near 1.9%.⁵ That is not a distressed old-economy Japanese industrial anymore; the re-rating has substantially happened. Note also that the outline framing of NGK as a sub-12x P/E value stock reflects a pre-rally valuation and no longer describes the security. The stock's own 52-week range — from ¥2,174.50 to ¥7,894 — is a reminder that this is now a high-beta expression of the AI capex trade, not a defensive holding.⁵

The activist stress test

What would a sceptical activist push on? Portfolio complexity, first: NGK still runs a metal-related business, a ceramic packages business it is restructuring, an industrial process line, insulators, energy plant engineering and radioactive waste treatment systems.¹ A focused argument would be that the semiconductor business is worth more separated from a declining combustion franchise and a grab-bag of small industrials, and that management's own portfolio matrix — which explicitly plots segments against a 12% hurdle — implies more disposals than have happened.¹ Second, disclosure: NGK provides good segment detail but limited English-language transparency on executive compensation and on how the divestiture screening process is actually applied. Third, the pace of returns: with ¥199.6 billion of cash and D/E of 0.3, the balance sheet could support materially more buyback than ¥33 billion without threatening the capex plan.¹ Management's answer would be that ¥450 billion of five-year capex is the better use of capital — a defensible position, and precisely the one that gets tested if semiconductor demand disappoints.

Two or three things to actually track

Cut through everything and there are three metrics that will determine whether this works.

One: Digital Society segment operating margin. Not revenue — margin. Revenue will follow the semiconductor cycle regardless. Margin is the read on whether NGK's ceramic components command genuine pricing power or merely ride volume. It expanded from 10.0% to 13.7% and is guided to 14.7%.¹ The question that matters is what it does in a down year.

Two: Environment segment operating income in absolute yen. The whole company's funding depends on this line holding near ¥67–68 billion while unit volumes decline.¹ The day this number starts falling meaningfully is the day the transformation clock starts running fast.

Three: group ROE against the 12% FY2030 target. It is the metric management has missed before, the one currency cannot rescue, and the single cleanest test of whether capital discipline is structural or reactive.¹

↑ Back to Top

X. Risk Radar

Risks are only worth listing when you can state the mechanism. Here are the ones with mechanisms.

Structural EV displacement of the Environment franchise. The transmission is direct: fewer combustion and hybrid vehicles produced means fewer honeycomb substrates, filters and sensors shipped. Content-per-vehicle increases from Euro 7, US Tier 4 and China 7 offset this for a period, and NGK's own reduction of its 2030 non-combustion share estimate to 30% extends the runway.¹ But there is no version of the long-term future in which combustion ceramics volumes are higher than today, and this segment funds the group. Watch the interaction: a fast EV re-acceleration would compress the window for Digital Society to reach scale.

Semiconductor capital expenditure cyclicality. Digital Society's earnings depend on tool makers' order books, which depend on fab construction decisions, which are notoriously prone to air pockets. The specific NGK exposure is that a downturn would coincide with rising depreciation from the ¥250 billion segment capex programme, so operating leverage that works beautifully on the way up works brutally on the way down.¹ This is the risk most likely to surprise investors who bought the stock on the last twelve months of results.

China, on both sides of the business. China is simultaneously a major automotive ceramics end market — with local honeycomb producers scaling and China VI/China 7 specifications set by Chinese regulators — and a semiconductor supply chain exposed to export controls.¹²¹ The mechanism on the automotive side is share loss to domestic competitors in the world's largest vehicle market; on the semiconductor side, restrictions on tool shipments to Chinese fabs reduce component demand indirectly, without NGK having any control over the policy.

Restructuring execution. The NAS wind-down is not finished on the day of the board resolution. NGK continues to service existing customer installations, which means warranty obligations, spare parts and field engineering for years on a product it no longer makes.⁸ Employees must be redeployed, the Chita insulator plant closure must be completed, and the ceramic packages restructuring within Digital Society must actually deliver the loss reduction management has built into FY2026 guidance.¹ Further charges beyond the ¥20 billion already taken would undercut the "one-and-done" framing.

Compliance in a concentrated industry. The 2015 case is a decade old, but the structural condition that enabled it — a handful of global suppliers selling qualified components to a handful of global buyers, with long negotiation cycles and price visibility — has not changed. This is a permanent feature of NGK's markets, not a resolved historical episode.

Input cost and energy. NGK fires kilns, which makes it structurally energy-intensive. Management explicitly built ¥2.0 billion of additional energy cost into FY2026 guidance because of crude price pressure linked to the Middle East, and flagged that no impact from broader market deterioration has been reflected in the forecast.¹ That last disclosure is honest and also a warning: the guidance assumes the geopolitical situation does not worsen.

Currency. FY2025's target beats were substantially yen-driven, at ¥151 to the dollar against a plan of ¥100.¹ With roughly 70% of sales overseas, meaningful yen strength would compress reported results without anything changing operationally.²

↑ Back to Top

XI. Durable Business & Investing Lessons

There are four things a long-term investor can take from NGK that generalise well beyond one Nagoya ceramics maker.

A deep technical competency is a renewable asset, but each deployment has a finite life. NGK has redeployed the same core capability — formulate, shape, fire, and control a ceramic through extreme conditions — into electrical insulators, spark plugs, catalyst substrates, particulate filters, exhaust sensors, sodium-sulfur batteries, hard disk components, and semiconductor process hardware. That is remarkable capital-efficiency of knowledge across a century. It is also true that insulators peaked, spark plugs left the building in 1936, NAS died, and substrates are on the clock. The competency renews; the individual markets do not. The investable question is never "does this company have great technology" but "is the current deployment early or late in its life, and is the next one visible yet." For NGK today, one deployment is late, one is early, and the gap between them is the entire investment case.

Cutting a prestige business is organisationally hard, and the delay is measurable. NGK's own disclosure states that ROE stagnated partly because of delays in exiting unprofitable businesses.¹ Companies rarely quantify their own procrastination that clearly. The lesson is to watch not whether a company eventually does the right thing, but how long it takes after the evidence is unambiguous — and to notice when the trigger is external. NAS was cut when BASF walked, not when lithium-ion won.³⁸ The institutional response to a formal screening process — hurdle rates, ROIC and free-cash-flow gates, a defined divestiture review — is the mechanism that shortens that lag, and it is worth more than any individual disposal.¹

A rebrand is usually a lagging indicator, and that makes it useful. The instinct is to dismiss corporate name changes as packaging. The more precise reading is to ask what the change is ratifying. When a company renames itself before the strategy has produced results, it is buying narrative time. When it renames itself after — after the growth segment nearly doubled profit, after the loss-maker was closed, after a record year — it is telling you the board believes the transformation has passed a point of no return. That is a real signal about management's own confidence, delivered at some cost.

External governance pressure works, and investors should price it accordingly. Japan's market-wide move — average PBR from 1.1 to 1.4 and ROE from 8.4% to 9.0% in three years following the TSE's 2023 request — is one of the cleaner natural experiments in corporate governance.¹⁸ NGK's dividend upgrade, escalating buybacks with cancellation, hurdle-rate increase and portfolio screening all arrived within that window, after decades of talking about ROE without delivering it. For investors, the useful generalisation is that reform-driven improvements are real but not proprietary: everyone gets them at once, which means the re-rating is a beta event, not an alpha one. The alpha comes from whether the company keeps going once the pressure normalises.

↑ Back to Top

XII. Epilogue

As of the April 2026 results and the name change that took effect three weeks earlier, NGK Corporation stands in a genuinely different place than it did five years ago. Records on sales, operating income and ordinary income, achieved while absorbing ¥20 billion of restructuring charges.¹⁴ A Digital Society business that grew from ¥171.6 billion to ¥205.4 billion of sales and is guided to ¥245.0 billion, with profitability improving faster than revenue.¹ A grid-storage chapter formally closed after four decades.³ A dividend rising 33% and a buyback of up to ¥33 billion with the shares to be cancelled.¹ And, for the first time, a strategic plan in which the fastest-growing and most profitable business is the one that has nothing to do with an engine.

What the company has not yet demonstrated is durability. The AI semiconductor upcycle has been generous to every supplier attached to it, and NGK's twelve-month share performance reflects a market that has priced the generosity as though it continues.⁵ The Environment business is being harvested competently, but competent harvesting of a declining asset is a different investment proposition from growth, and management's own guidance — volumes down, revenue up, profit flat — describes it exactly.¹

Three things will settle the argument over the next few years. Whether Digital Society's profitability holds through a semiconductor downcycle, which is the only real test of whether NGK owns pricing power or merely capacity. Whether Environment's absolute operating profit holds near current levels as combustion volumes decline, since that line funds the ¥450 billion five-year investment programme.¹ And whether ROE actually converges toward the 12% FY2030 target, which requires both of the above plus continued discipline on the equity base.¹

There is also a smaller, more human thing to watch. A company that changes its name is making a claim about identity. The interesting question is whether NGK Corporation starts reporting like the company its new name describes — whether the segment disclosure, the capital allocation, the executive incentives and the compliance narrative all migrate toward the digital-and-carbon-neutral company on the letterhead, or whether the letterhead simply sits atop an organisation whose centre of gravity, culture and cash flow still live in Nagoya, in a kiln, making parts for engines.

↑ Back to Top

XIII. Outro & Links

For readers who want to work from primary materials rather than summaries, the FY2025 results presentation dated April 30, 2026 is the single most information-dense document in the file: it contains the segment forecasts, the product-line revenue splits, the capital cost estimates, the hurdle-rate change, the dividend and buyback details and the FY2030 targets, all in one place.¹ The supplementary data pack accompanying it carries the historical segment series.²⁰ The NAS discontinuation notice of October 31, 2025 is short and worth reading in the original for its restraint.¹⁵ The Justice Department's 2015 press release and NGK's own plea disclosure remain the cleanest account of the cartel episode.¹⁰¹¹ And the NGK Group Vision "Road to 2050" sets out the long-range framework — carbon-neutrality and digital-society-linked products at 50% of sales by 2030 and 80% by 2050 — against which everything above should be measured.²¹ Prior-year presentations in the IR library are the best way to check whether the current narrative matches what management was saying two and three years ago.²²

References

1. FY2025 Financial Results Presentation (year ended March 31, 2026) — NGK Corporation, 2026-04-30 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. Message from the President — NGK Corporation ↩↩↩↩↩

3. Japan's NGK discontinues manufacturing of sodium-sulfur batteries — pv magazine, 2025-11-05 ↩↩↩↩↩↩↩

4. NGK Insulators, Ltd. Full Year Income Advances — RTTNews, 2026-04-30 ↩↩↩

5. NGK Corporation (TYO:5333) stock quote and statistics — StockAnalysis ↩↩↩↩

6. History — NGK Corporation ↩↩↩↩↩↩↩

7. NAS Batteries (Sales Discontinued) — NGK Corporation ↩↩↩

8. NGK pulls plug on world's second-most-deployed grid storage battery technology after BASF exit — Energy-Storage.News, 2025-11-05 ↩↩↩↩↩↩↩

9. Cause of NAS Battery Fire Incident, Safety Enhancement Measures and Resumption of Operations — NGK, 2012-04-25 ↩↩

10. NGK Insulators Ltd. to Pay $65.3 Million for Fixing Prices on Auto Parts — U.S. Department of Justice, 2015-09-03 ↩↩↩↩↩↩

11. Plea Agreement with U.S. Department of Justice concerning ceramic substrates for catalytic converters — NGK, 2015-09-04 ↩↩↩

12. Ceramic Honeycomb Substrate Market, Global Outlook and Forecast 2025-2032 — 24chemicalresearch, 2025-07-31 ↩↩↩↩↩

13. NGK to Triple Production Capacity for Translucent Alumina Wafer — NGK, 2025-11-14 ↩↩

14. Electrostatic Chuck for Semiconductor Process Market Outlook 2026-2034 — Intel Market Research, 2026-07-25 ↩↩

15. Notice Regarding Discontinuation of Manufacturing and Sales Activities of NAS Batteries — NGK, 2025-10-31 ↩↩

16. Directors and Corporate Officers — NGK Corporation ↩

17. Financial and Capital Strategies for Supporting Transformations (CFO interview) — NGK, 2022 ↩

18. Tokyo Stock Exchange Initiative on Cost of Capital and Stock Price Conscious Management — Harvard Law School Forum on Corporate Governance, 2025-10-21 ↩↩↩

19. NGK Insulators Group will change its name to NGK Corporation on April 1, 2026 — NGK Berylco ↩

20. FY2025 Results Supplementary Data — NGK Corporation, 2026-04-30 ↩

21. NGK Group Vision: Road to 2050 — NGK Corporation ↩↩

22. IR Presentation Library — NGK Corporation ↩