Disco: A Long-Term Business Owner's View

There are three key uncertainties. First, the market is currently willing to pay an extremely high multiple, which is essentially a bet that demand for AI, advanced packaging, and high-value-added products stays strong for a long time; once the cycle normalizes, valuation compression alone could swallow years of fundamental progress. Second, reported free cash flow has been highly volatile in recent years, and in FY2025 it even turned negative because of the timing of term deposits and investments, so estimating the company's "true distributable cash flow" requires adjustment, and the conclusion is sensitive to assumptions. Third, although the industry has high barriers, it ultimately belongs to the semiconductor equipment chain, and the company itself explicitly provides forward guidance for only a single quarter, because customer investment appetite swings sharply and full-year visibility is limited.

The One-Sentence Judgment

If you ask "is Disco an excellent company worth following for the long term," my answer is yes; if you ask "at today's price, does it meet a Buffett-style conservative buy," my answer is no.

Business Logic and Industry Position

Understanding the Business: How Does This Company Actually Make Money?

Disco's core business is supplying high-precision cutting, thinning, and polishing equipment for semiconductor and electronic-component manufacturing, along with the matching blades, grinding wheels, and polishing wheels consumables, and providing maintenance, training, application testing, and process support. The company has long summarized its business as "Kiru, Kezuru, Migaku" — that is, cut, grind, polish; Reuters' company profile likewise describes it as a maker of semiconductor manufacturing equipment and precision processing tools, with products including dicing saws, laser saws, grinders, and polishers. In other words, this is not a business that ends once a machine is sold; it is an "equipment + tools + know-how + service" combination in which the installed base keeps pulling demand for consumables, maintenance, and process services.

Its customers are mainly semiconductor manufacturers, electronic-component makers, and customers along the assembly/test and OSAT chain. The company's "Business" page points clearly to semiconductor manufacturers and electronic-component makers as customers; the FY2025 4Q briefing directly noted stronger dicer shipments to ICs and OSATs. On the revenue mix, the FY2025 4Q earnings briefing breaks quarterly sales into Precision Processing Equipment, Precision Processing Tools (Consumables), Maintenance Parts, and Other, showing that revenue comes both from one-time equipment sales and from higher-frequency consumables and maintenance. For a long-term owner this matters a great deal, because it means the business does not simply "feed on capex" but also partly "feeds on the utilization of the installed base."

Revenue is not fully recurring or stable, but it is highly understandable. It is not as smooth year to year as consumer goods, nor is it the "one-off sale" of a pure project-based equipment maker. Equipment revenue swings with customer capex, acceptance schedules, and application cycles, while consumables and maintenance revenue are highly correlated with customer equipment utilization, producing a "semi-recurring but still cyclical" revenue structure. The FY2025 briefing noted that consumables sales hit a record high on the back of customer equipment utilization; meanwhile, in its FY2024 annual report the company stressed that because the semiconductor market fluctuates sharply and rapidly, it discloses earnings forecasts only one quarter ahead. That alone tells you this is a business you can understand, but not one whose operating figures two or three years out can be seen clearly.

On cost structure, the company's high gross margin comes from high-value-added products, process know-how, the consumables mix, and service support, but it also continually spends heavily on R&D and personnel. Full-year FY2025 sales were about ¥436.9B, operating profit about ¥185.0B, and the operating margin about 42.3%; the company also said that the rise in SG&A came mainly from personnel and R&D spending. For a manufacturer, such a margin is highly unusual and shows that it is not competing on cheap scale but selling "precision, yield, reliability, and a complete solution."

On dependence, what I can confirm is that the company is highly sensitive to semiconductor-industry conditions and customer capex and highly exposed to overseas markets; in FY2025 4Q, overseas sales reached 87.6% of the total. I cannot, from the materials reviewed here, confirm the latest top-five-customer concentration, and I did not find recent English-language materials with sufficient disclosure on single-customer concentration, so I can only conservatively note that this item "needs verification against the latest securities report."

If the stock market closed for five years, would I want to own this business? I would want to own the business, but not buy it outright at today's price. The business itself is transparent, its economic logic is clear, and the moat is visible; the problem is that the price has already prepaid a great deal of good news.

Business Understandability Score: 4/5.

This is not complex financial engineering, nor an inscrutable internet platform; the technical details run deep, but the money-making logic is clear. What truly deserves respect is not "a hard-to-understand business" but "a steep cycle and a steep valuation."

Industry and Competitive Landscape

Disco operates not in a "stable, volatility-free" ordinary industrial-equipment industry, but in a high-barrier, technology-intensive, yet highly cyclical niche of semiconductor equipment and precision processing. In its mid-2025 outlook, SEMI projected that the Assembly & Packaging equipment market would grow about 8% to $5.4B in 2025 and grow another 15% in 2026; SEMI also identified AI, HPC, advanced packaging, HBM, and chiplet architectures as important long-term drivers of front-end and back-end equipment investment. WSTS likewise said the global semiconductor market would keep growing in 2026. So the long-term direction of demand is a tailwind, but the path there is not smooth.

Long-term industry demand is broadly on a stable upward trajectory, because there are more chips, packaging is more complex, wafers are thinner, and yield requirements are higher — all of which raise the importance of the "cut, grind, polish" stage. In its FY2025 earnings briefing, the company stated clearly that FY2026 customer investment demand still centers on generative-AI-related applications, and it expects the adoption and volume production of advanced packaging technology to advance further; but it also cautioned that shifts in non-AI application demand still warrant close watching. In other words, the tailwind exists, but it does not mean there is no risk.

This industry is easily affected by new technology cycles, but not easily disrupted. The reason is that disruption is not just proposing a new process; it must hold up simultaneously across precision, speed, yield, material compatibility, long-term stability, and customer validation. Disco itself does not sit still on a single legacy technology — its products already span blade dicing, laser saw, grinding, polishing, surface planer, and waterjet saw, showing that the company is actively incorporating new approaches that could replace older processes rather than passively waiting to be displaced.

On competition, the closest listed comparable is Tokyo Seimitsu; a broader equipment comparison would include Tokyo Electron, while among the high-valuation AI-chain names, Advantest also offers a reference for "how high a valuation the market is willing to pay for a high-quality semiconductor asset." It must be stressed that these companies do not have entirely the same business focus, so relative valuation can only assist in gauging price, not replace an independent judgment of Disco's business quality.

On an industry-quality basis, my conclusion is not as simple as "a good company in a good industry," but rather: an excellent company in a high-growth, high-barrier, but strongly cyclical industry. The most common mistake with such companies is not misreading the business model, but buying a good business at too high a price during a boom.

Industry Attractiveness Score: 4/5.

Long-term demand is strong, the technical barriers are high, and advanced packaging is a tailwind; but the earnings volatility and customer investment cycle mean it is not a perfect industry.

Moat and Management Quality

Moat Analysis

Disco's moat rests not on brand advertising but on process precision, the equipment–consumables pairing, application know-how, a global service system, and organizational capability.

There is a brand advantage, but more precisely it is a technical brand and process credibility. Customers are not buying the "halo of a famous name" but higher yields, fewer chip-outs, and more stable processing of thinner wafers and more complex materials. Reuters, the company's website, and its annual corporate reports all position the products at the key process steps of semiconductor manufacturing.

The cost advantage is not the shallow "low-cost manufacturing" but a superior total cost of ownership. If a customer uses worse equipment or consumables, it may pay a higher price in blade wear, yield, processing speed, maintainability, and downtime. The company's 2020 corporate report stresses clearly that DISCO's core is not just the equipment and consumables themselves, but also application labs, processing-condition recommendations, and global after-sales and training. This means what customers truly buy is "the result."

The scale advantage lies in the installed base + service network + data-feedback loop. The company has global application labs and an after-sales support system; its Trouble Tracking System shares global fault information in real time for cross-organization troubleshooting and improvement. Such a system means its scale is not as simple as "opening a few more factories," but rather distilling the experience of customer sites worldwide into the speed and quality of problem-solving. For precision processing equipment, this accumulation naturally translates into higher customer stickiness.

The network effect is a weak network effect, not a platform-type strong network effect. The more customers there are, the more problem samples the company gathers, the richer its application parameters, and the more mature its service system, which increases its appeal to new customers; but it is not the kind of social network where "value rises exponentially for all users as the user base grows." So its moat is more of an "experience and validation barrier" than a classic platform moat.

Switching costs are very high. Once a customer brings a given piece of equipment, consumable, and set of process parameters into a volume production line, switching suppliers later involves re-validation, yield risk, engineering time costs, and the opportunity cost of line downtime. By running application tests and recommending optimal processing parameters, the company deepens this embedding. Consumables reinforce the mechanism: the larger the installed base, the more likely later consumables and maintenance purchases continue along the existing system.

The channel advantage shows up in direct technical and service channels to the customer, not in distributor stocking. For this kind of complex equipment, what truly adds value is pre-sale trial cuts, after-sales service, training systems, fault response, and process collaboration, rather than a plain sales channel. The company's emphasis on training, application support, and after-sales service comes through very clearly in its corporate reports.

Patent, technology, and organizational barriers are significant, but the materials reviewed here do not provide a consistent, up-to-date count of patents, so I will not fabricate a specific figure. What can be confirmed with high confidence is that this business requires long-term, integrated technical accumulation across mechanics, electrical engineering, physics, chemistry, and IT, and the company keeps adding to its R&D and capacity. The FY2026 1Q outlook page states explicitly that the company will continue to strengthen R&D and production capacity, advancing facilities such as a new building at the Haneda R&D Center and a new plant at Hiroshima Works.

The data advantage shows up more in process parameters, fault cases, application trials, and customer-site experience than in internet-style user data. For an industrial company of this kind, such data is more valuable and harder to replicate.

Corporate culture and operating capability are, in my view, Disco's most easily underestimated moat. Its public materials repeatedly emphasize DISCO VALUES, Will Accounting, PIM activities, and RORA management; more importantly, these are not empty slogans but are tied to incentive mechanisms — the company even notes that the four-year RORA target is linked to employee bonuses. Management also states explicitly that high market share and high profitability are the results of measures to "strengthen the company," not goals in themselves. Such statements are very rare and very much resemble running the business from a genuine owner's perspective.

Is This Moat Widening, Stable, or Narrowing?

My judgment is: overall it is still widening, but the valuation is now far ahead of the pace at which the moat widens. Trends such as advanced packaging, AI, high-bandwidth memory, and chiplets favor this kind of high-precision back-end and thinning/dicing capability on the demand side; the company is also continuing to invest in R&D, facilities, and organizational systems. The problem is that the valuation the market assigns already implies "the moat keeps widening with almost no missteps." A company can be outstanding, yet a stock need not be bought at any price.

Difficulty of Replication

For a competitor to replicate Disco, it is not enough to spend some capex and buy a few machines; it must catch up simultaneously across equipment performance, consumables development, process parameters, global service, customer validation, and organizational culture. In time, this would likely take years; in capital, it is not just fixed assets but also the cost of talent and customer validation.

Can It Raise Prices in Inflation, Stay Profitable in Downturns, and Are the Past High Margins Structural?

Looking at the history, the company's margins have risen sharply and held high, but not entirely on the back of cyclical tailwinds. The FY2020 operating margin (for the year ended March 2021) was already about 29.0%, and from FY2022 to FY2025 it climbed further into roughly the 39%–42% range; even in the weaker environment of FY2019, the operating margin on the company's basis was still 25.8%, with an ordinary profit margin of 27.2%. This shows the margin reflects both a cyclical tailwind and very strong structural advantages. My judgment is: most of the high margin has a structural foundation, but at this level the ultra-high valuation is still betting on the cyclical tailwind continuing.

Moat Strength Score: 4.5/5.

I am willing to give it a very high score; the 0.5 point deducted is not because the moat is weak, but because it does sit in a capex-cycle industry and cannot be as naturally smooth as Coca-Cola.

Management and Capital Allocation

What I respect most about management is not "telling many beautiful stories," but having put in writing a set of verifiable, trackable, and constrained capital-allocation principles: not targeting sales scale and global share, but targeting business quality; using four-year ordinary profit margin and four-year RORA rather than single-quarter or single-year metrics to assess operations; keeping capital inside the company to weather volatility, fund R&D, and acquire technology resources; returning capital to shareholders primarily through dividends, while institutionalizing the return of part of "excess cash" to shareholders. Such a framework comes very close to long-term owner thinking.

On alignment of interests, the 2026 shareholders' meeting notice shows that President and Representative Executive Officer Kazuma Sekiya holds 2,101,400 shares; estimated against the company's current total share count of about 108 million shares, that is roughly 1.9% of shares outstanding. This proportion falls short of control, but for a professional manager it already represents a meaningful alignment of interests. The company uses a three-committee structure and in 2026 elected 10 directors, 7 of them external, with an overall governance framework that is reasonably mature.

The capital-allocation record is broadly rational. The company has long held net cash, with cash and deposits of about ¥284.6B and total liabilities of about ¥155.3B at the end of FY2025; the dividend policy is clear, and the company added to dividends in high-cash years. The company also states explicitly that buybacks are only a supplementary tool, not the primary form of return; the change in shares outstanding has been very small in recent years, indicating that dilution from equity incentives or options is not aggressive. Based on the materials I have seen, the company looks more like a "high-dividend, low-buyback, strong-reinvestment, low-leverage" capital-allocation style.

On M&A, in the materials reviewed here I found no evidence that large acquisitions have been a main growth driver in recent years; the company's dividend formula reserves room for future technology-resource acquisition, patent purchases, and potential investment/M&A, but current growth looks driven more by organic competitiveness than by acquisitions. Because I have not item-by-item reviewed the notes to the latest securities report, I retain a cautious "needs additional information" stance on this point.

Management and Capital Allocation Score: 4/5.

I rate its long-term orientation and discipline very highly; the deduction is mainly because the industry it operates in is naturally volatile, no management can fully eliminate cyclicality, and the long-term returns from its recent heavy facility investment still need time to verify.

Financial Quality and Owner Earnings

Financial Quality Analysis

Start with the most important conclusion: Disco's profit is largely real cash profit, not paper profit; but reported free cash flow is heavily affected by the pace of investment and the conventions of cash management. Over fiscal years 2021–2026 (ended March), revenue grew from about ¥182.9B to about ¥436.9B, a roughly 19% five-year CAGR; operating profit and net profit compounded at about 28% over the same period, clearly faster than revenue, showing that margin expansion and operating leverage together drove value creation. Estimated from official data in this report, the FY2025 operating margin was about 42.3% and the net margin about 31.0%; the FY2020 operating margin was already about 29.0%, and even in the weaker environment of FY2019 it was still 25.8%. This is profit resilience that ordinary cyclical companies cannot achieve.

All amounts in the table are in billions of yen; ROE and ROIC are estimated in this report on a public-reporting basis, with ROIC using an approximate "after-tax operating profit / average invested capital (shareholders' equity + liabilities − cash and deposits)" method, intended only as a directional reference and not suitable for direct cross-comparison with peers' disclosed values on a non-standardized basis. Revenue, profit, assets, equity, operating cash flow, investing cash flow, and FY2026 free cash flow in the table are drawn mainly from the company's FY2020–FY2025 annual results materials; the March 2026 figures come from the FY2025 4Q briefing and the cash-flow/balance-sheet summary, while the 2024–2025 figures come from the annual results announcements.

Cash-flow quality is quite good. Across the six fiscal years ended March 2021–2026, cumulative operating cash flow was about ¥573.7B, slightly above cumulative net profit of about ¥531.8B, showing that earnings were not built on receivables or inventory. Working capital does tie up cash in high-growth years: the FY2025 operating cash-flow statement shows receivables up about ¥11.0B, inventory up about ¥0.2B, and payables down about ¥10.7B, which together with higher tax payments markedly compressed reported free cash flow. In other words, the profit is real cash, but free cash flow gets "broken up" by growth and the timing of cash management.

What most needs explaining is the negative free cash flow in FY2025. The company disclosed FY2025 operating cash flow of about ¥133.5B, investing cash flow of about -¥135.8B, and free cash flow of about -¥2.2B; but within that, "Others" was about -¥100.6B, mainly term deposits, not all genuine operating capex; PP&E purchases over the same period were about ¥35.1B. So mechanically reading the FY2025 -¥2.2B as "true earning power has collapsed" is wrong. It reflects more that the company parked a large amount of cash in term deposits while continuing to advance plant and R&D facility investment. For a long-term owner, this should be unpacked rather than judged on a single FCF number.

The balance sheet is very robust. At the end of FY2025, total assets were about ¥743.4B, total liabilities about ¥155.3B, net assets about ¥588.1B, and the equity ratio about 78.9%; the company has long held net cash, with very low non-current liabilities. On this structure, net debt/EBITDA is in fact negative, and the interest-coverage ratio is not a major constraint, because the company barely earns by leverage. The real risk is not financial leverage but a cyclical downturn after buying at a high valuation.

On inventory, receivables, and payables, the recent picture shows the "normal stretching of a high-growth company." From FY2023 to FY2025, both inventory and receivables rose, driven by capacity, deliveries, and demand expansion; but there is no sign of runaway receivables, abnormal bad debts, or cash flow chronically diverging from profit — the kinds of typical accounting red flags. FY2024 also saw an impairment/special loss of about ¥7.5B related to the rebuilding of the Haneda R&D Center, which instead reminds us that the company has not deliberately hidden every unfavorable item in its accounting. So far, I have seen no obvious signs of financial fraud, aggressive accounting, or earnings manipulation.

On share count, the change has been very small. On a post-split basis, shares outstanding went from about 108.18 million (March 2021) to about 108.42 million (March 2025), an increase of less than 0.3%; this means the company has not diluted shareholders through large issuance or intensive equity incentives. Dividends, meanwhile, have kept rising: on a post-split basis, the annual dividend rose from about ¥225.7/share for the year ended March 2021 to about ¥505/share for the year ended March 2026. This is a rather handsome shareholder-return curve.

A Special Judgment

The profit looks more like real cash profit than an accounting illusion. Growth does require capital investment, but it is not a "the more it grows, the more cash-starved it gets" model; more precisely, it is "incremental business is highly profitable, but cash collection appears volatile because of project acceptance, inventory, capacity construction, and term-deposit allocation." In a downturn, the company still has ample ability to survive, because it carries long-term net cash, exceptionally high gross and operating margins, and a conservative capital structure. What should truly worry you is not survival risk, but that the entry price has prepaid for the next decade of good news.

Owner Earnings Analysis

On an "owner earnings" basis, FY2025 accounting net profit was about ¥135.5B. The main non-cash expense to add back is depreciation and amortization of about ¥14.8B; what needs to be subtracted is maintenance capex and maintenance working-capital usage. The difficulty is that the official reports do not directly disclose "maintenance capex," and FY2025 investing cash flow was dragged down by about ¥100B of term deposits, so a subjective split is required.

My conservative approach is to estimate FY2025 maintenance capex at about ¥18B, above depreciation, reflecting the company's need to keep updating equipment and process capability; and to estimate maintenance working-capital usage in the ¥10B–¥20B range, acknowledging that the company must maintain a high level of inventory, delivery, and service capability. On that basis, FY2025 conservative owner earnings come to roughly ¥110B–¥125B, with a midpoint of about ¥120B, equating to roughly ¥1,100 per share. In other words, the current share price of about ¥71,520 equates to roughly 62–71x owner earnings. For even the finest industrial company, this multiple implies extremely high expectations.

This also explains why I am unwilling to focus only on FCF under GAAP/JP-GAAP. Treating the FY2025 reported free cash flow of -¥2.2B as "true cash-generating ability" would be far too pessimistic; treating net profit of ¥135.5B as entirely "freely distributable" would be far too optimistic. The truly reasonable anchor is the owner-earnings range above.

Valuation and Margin of Safety

Intrinsic Value Estimate

I value the company three ways, and the conclusions are highly consistent: the company is excellent, but the current price is unfriendly to conservative investors.

Method One: Owner-Earnings Discount

My three-scenario assumptions are as follows. To avoid misleading anyone, the valuations below are all estimates from this report's model, not company guidance and not the sell-side consensus.

This result shows that, at today's price of about ¥71,520, the market not only requires Disco to remain a good company but also requires it to deliver "high growth + high margins + high-quality capital returns" over a long period in the coming decade, with the discount requirement kept low. For "balanced-to-conservative" capital, this is not a margin of safety but a very narrow room for error.

Based on the model above, I set out the following value ranges:

• Conservative intrinsic value range: ¥19,000–¥25,000/share

• Fair intrinsic value range: ¥30,000–¥42,000/share

• Optimistic intrinsic value range: ¥55,000–¥68,000/share

On that basis, the current price relative to:

• the conservative range: a premium of about 186%–276%

• the fair range: a premium of about 70%–138%

• the optimistic range: still roughly a premium of 5%–30%

If you insist on a Buffett-style margin of safety, I believe you need at least about a 25% discount to fair value before you can speak of an "ideal entry."

So my price-band judgment is:

• Ideal buy range: ¥22,000–¥31,000/share

• Acceptable hold range: ¥30,000–¥50,000/share

• Clearly overvalued range: broadly above ¥60,000/share

This is not to say the share price cannot go higher in the short term; only that, from the standpoint of acquiring a business for the long term, this price band already makes it hard for me to be satisfied.

Method Two: Relative Valuation

At the current price, Disco's valuation is significantly higher than most comparable Japanese equipment companies on both a trailing and a consensus basis. Google Finance shows that on June 9, 2026 its trailing P/E was about 57.4x and market cap about ¥7.76 trillion; measured against FY2025 year-end net assets in this report, the current P/B is about 13.2x. On MarketScreener's FY2026 consensus basis, Disco trades at about 49.0x P/E, 11.3x P/B, and 31.8x EV/EBITDA. Tokyo Electron corresponds to about 29.7x P/E, 8.28x P/B, and 23.3x EV/EBITDA; Tokyo Seimitsu to about 21.7x P/E, 2.81x P/B, and 17.3x EV/EBITDA. Even accounting for Disco's stronger earnings quality and deeper niche moat, this premium is already very high.

Note: peer valuations in the table come mainly from MarketScreener's consensus basis, suitable for a side-by-side comparison of "how high the expectations the market assigns," but not for mechanically treating different business models as identical comparables; Disco's current trailing valuation also references Google Finance. The FY2026E EV/FCF for Disco shows an extreme value precisely because its reported FCF is heavily affected by cash management and the pace of investment, which further shows that a single multiple cannot be applied mechanically.

Against a broad index, the valuation gap is also clear. The official Nikkei index history page shows that the Nikkei 225's market-cap-weighted PER was about 17.47x on June 8, 2026; Disco's trailing P/E on June 9 was about 57.4x. This means that buying Disco today is not buying "an ordinary excellent company among Japanese stocks," but buying "one of the most expensive classes of high-quality semiconductor assets in the Japanese market."

Method Three: Asset or Liquidation Value

For Disco, the asset method is more of a floor than the core of value. At the end of FY2025, total assets were about ¥743.4B, net assets about ¥588.1B, and cash and deposits about ¥284.6B; even on a rough net-cash estimate (cash and deposits minus total liabilities), net cash is only on the order of about ¥129B, equating to a little over a thousand yen per share. In other words, today's share price is barely buying assets; it is paying for many years of excess earning power to come. The balance sheet gives you safety, but it cannot give the current price much of an "asset floor" to lean on.

Margin of Safety

My answer is clear: the current price does not offer an adequate margin of safety.

The most fragile assumption in the valuation is that "high growth, high margins, and a high valuation can all hold up for a long time." If growth falls short, returns will be revised down quickly; if margins slip from around 42% back toward the upper end of the historical 30%–35% range, the valuation will quickly become untenable; if the multiple compresses from 50–60x to 25–35x, then even if profits keep growing, shareholders may endure a long stretch of low returns or even permanent capital loss. For a stock already highly sought after by the market, these are not paper risks but the most realistic ones.

This is the textbook case of a "good company but bad price." Conservative investors should acknowledge that if you buy now, what truly makes you money is not that you found an undervalued quality business, but that you bet correctly on the continued strength of AI capex and the advanced-packaging boom over the next several years. That is not something value investing cannot do, but it has clearly drifted away from the starting point of "buying a good company with an adequate margin of safety."

Risk Comparison and Execution Checklist

Risks and the Bear Case

The most important risk is not short-term price volatility, but the several sources of permanent capital loss:

First, overvaluation risk. This may be the single biggest risk at present. The 10-year JGB yield had risen to about 2.74% on June 9, 2026, meaning the risk-free rate is not low; in such a rate environment, paying 50–60x earnings for a single semiconductor-equipment leader naturally lowers the room for error.

Second, cyclical risk. The company itself admits customer investment appetite swings sharply, which is why it gives only one-quarter guidance. A long-term upward industry does not mean every year is up, still less that the current high level of capacity and demand will extend smoothly.

Third, risk of technology substitution or shifts in product mix. Advanced packaging and new processes such as laser/plasma do not necessarily weaken Disco, and it is actively positioning for them; but if certain mainstream process roadmaps in the future reduce the economic appeal of the existing equipment–consumables combination, the company's pricing power could be marginally eroded.

Fourth, customer and application concentration risk. Although the company covers the globe with a high share of overseas sales, recent incremental demand is clearly more concentrated in high-activity applications such as AI, ICs, and advanced packaging; once investment in these niches cools, both shipments and acceptance pacing could be affected.

Fifth, supply-chain, geopolitical, and regulatory risk. SEMI has already listed U.S. tariffs, export controls, and regional supply-chain restructuring as important uncertainties for the equipment industry; in its FY2026 outlook the company itself specifically cautions about the impact of the international situation on the operating environment.

Sixth, currency risk. The company's FY2026 1Q guidance uses an assumption of ¥157 to the U.S. dollar and discloses an annualized USD sensitivity of about ¥1.7B. A high share of overseas sales means currency can amplify profit but can also compress earnings when the direction reverses.

The Strongest Counterargument

The strongest counterargument is actually simple: Disco may genuinely be a superb company, but what you buy today is more "superb expectations under market sentiment" than "undervalued business worth." Those who are bearish usually see not a deteriorating company but "a valuation that is too demanding of the future." If AI capex normalizes somewhat over the next three to five years, the advanced-packaging pace falls short of what the market imagines, or margins return to a slightly lower level, then even if the company remains first-rate, today's buyers may not earn a decent return.

Which Facts Would Overturn My Current Judgment

The facts that would overturn my "watch rather than buy" judgment are mainly these:

• The company steadily lifts owner earnings over the coming years to well above my current estimate, for example approaching and durably exceeding ¥150B–¥180B;

• The high margin proves not to be a phase but can be maintained over the long term at a larger revenue scale;

• The share of consumables, maintenance, and service revenue rises further, making cash flow more stable than I now judge;

• The market price falls back into a range that offers an adequate discount.

Conversely, if you already hold the stock, the following facts should trigger a reassessment:

• A sustained step-down in margins attributable to competition or process change;

• A clear deterioration in the four-year ordinary profit margin and RORA;

• High cash and high capex coexisting, but new capacity failing over the long term to translate into higher owner earnings;

• The installed base no longer effectively pulling consumables and maintenance;

• Management beginning to pursue scale rather than per-share value.

Comparison with Other Opportunities

Comparing Disco with Tokyo Seimitsu, Disco clearly has higher earnings quality and a deeper niche moat, and so earns a higher valuation; but from the standpoint of "buying today," Tokyo Seimitsu's valuation is more forgiving of mistakes. Comparing it with Tokyo Electron, Disco is more focused and possibly deeper, but Tokyo Electron's valuation is also clearly lower. Comparing it with the Nikkei 225 / a broad Japanese index, Disco could of course outperform over the long term, but you are taking single-track, single-company, single-high-valuation risk rather than diversification. Comparing it with the 10-year JGB at about 2.74% risk-free yield, Disco of course has higher long-term return potential, but only if you are willing to accept valuation drawdowns under extremely high expectations. For balanced-to-conservative capital, I believe it is not yet clearly superior to the two alternatives of "waiting for a better price" or "first holding a broad index."

If I could hold only five assets, at the current price I would not put it in the portfolio. But if the price falls significantly while the business quality holds, it would immediately move to the front of the candidate list.

Investment Checklist

Final Investment Conclusion

【Final Rating】 Watch

【One-Sentence Investment Thesis】 Disco is an exceptionally high-quality, deeply moated leader in semiconductor precision processing, but the current share price looks more like prepaying for years of ultra-optimistic growth than leaving a margin of safety for a long-term owner.

【Core Bull Case】

• Sits at the key "cut, grind, polish" step of semiconductors, with a high technical and process standing.

• Equipment, consumables, maintenance, and application support form a combined business model with strong customer stickiness.

• Historical margins, ROE, estimated ROIC, and net cash levels are all excellent.

• Management openly uses four-year operating metrics and a disciplined capital-allocation framework, lending credibility to its long-termism.

• AI and advanced-packaging trends are a medium-to-long-term tailwind for demand for its core processes.

【Core Bear Case】

• Both trailing and forward valuations are extremely high, with an insufficient margin of safety.

• The industry itself is strongly cyclical, and the company gives only single-quarter guidance.

• FY2025 reported free cash flow was negative, and judging true distributable cash flow is sensitive to assumptions.

• The current price largely bets that AI and advanced-packaging demand keep beating expectations.

• If the multiple reverts, shareholder returns could be poor even if the business stays excellent.

【Key Assumptions】

• High-frequency, high-value-added equipment and consumables demand keeps being pulled by AI, advanced packaging, and high-performance chips.

• The company maintains above-industry-average margins and returns on capital.

• Organizational capability, the service system, and the consumables pairing keep strengthening the moat.

• Capex ultimately converts into higher long-term owner earnings.

【Fair Buy Price】 A more conservative ideal buy range is about ¥22,000–¥31,000/share; if you have strong conviction in its moat, you might at most treat ¥30,000–¥42,000/share as a "discussable range." The basis is this report's conservative–neutral discount of owner earnings, not a guess at short-term market sentiment.

【Target Holding Period】 At a reasonable price, it suits a holding period of 10 years or more; at an unreasonable price, no length of holding can automatically fix the problem of overpaying.

【Expected Annualized Return】 On this report's rough scenario estimates of owner-earnings growth, dividends, and exit valuation over the next 10 years:

• Conservative scenario: about -3%/year to 0%/year

• Neutral scenario: about 2%/year to 4%/year

• Optimistic scenario: about 8%/year to 10%/year

This distribution of returns is not attractive enough for "balanced-to-conservative" capital, because its upside requires many good things to happen at once, while its downside scenario is not far-fetched.

【Maximum Loss Risk】 In the worst case, if AI capex and the advanced-packaging boom cool, margins return to a more ordinary equipment-company range, and the market valuation compresses back to 25–35x earnings, a decline of 40%–60% or even more from the current level relative to today is not unimaginable; the true permanent loss would come from "buying high + earning only mediocre returns over the long term."

【Tracking Metrics】 Please keep tracking the following metrics:

• Quarterly shipment figures

• Precision Processing Tools (consumables) growth and share

• Ordinary profit margin and operating margin

• RORA and ROE

• Divergence between operating cash flow and reported FCF

• PP&E purchases and returns from newly commissioned facilities

• Cash and term-deposit size, and execution of the supplementary-dividend formula

• Overseas sales mix and currency sensitivity

• Changes in the share of AI/advanced-packaging-related application revenue

• Competitors' technical catch-up in dicing / grinding / advanced packaging

【Signals That Trigger Reassessment】

• The four-year cumulative ordinary profit margin falls clearly out of the company's long-term target band;

• Consumables and maintenance revenue fail to keep pace with the installed base;

• Large capex occurs repeatedly, but owner earnings fall rather than rise;

• New process roadmaps weaken the economics of existing products;

• Management begins emphasizing scale and share while playing down asset efficiency and return discipline;

• The valuation falls back into a range with an adequate margin of safety, or conversely keeps expanding well away from fundamentals.

【Final Recommendation】 Said calmly and with restraint: put Disco on the high-quality watchlist, not on today's buy list. This is a company worthy of respect, worthy of long-term study, and even worth considering for a large position when it is cheaper; but buying at the current price looks more like prepaying a hefty deposit for flawless future execution than protecting capital with a margin of safety. For a long-term value investor, one of the hardest skills is not finding a good company, but being willing to do nothing when a good company is clearly not cheap.

Open Questions and Material Limitations

• This work prioritized the company's website, annual results materials, the shareholders' meeting notice, official management briefings, and a small amount of authoritative secondary financial data; the full text of the latest EDINET securities report was not reviewed item by item, so the top-five-customer concentration and finer segment detail still warrant additional confirmation.

• Peer ROIC lacks a consistent and fully comparable public basis, so the cross-comparison relies mainly on P/E, P/B, and EV/EBITDA.

• This report's owner earnings, ROIC, intrinsic value ranges, and expected returns are all model estimates, valuable for helping you recognize "how high the expectations are" rather than providing a falsely precise target price.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.