Business

Know the Business

Msscorps runs a microscope shop for the chip industry — a capital-heavy laboratory that gets paid per analytical report to dissect samples for foundries, IC design houses, and OSATs. Revenue tracks customer R&D budgets, not wafer starts, so demand is stickier than the chip cycle, but a single line item — equipment depreciation — controls whether the lab earns 40% gross margin or 20%. The market price assumes a FY26 margin snap-back from the FY24–25 capex wave; the whole question on the stock is whether the tools that the company has bought get utilised at the rates management has promised.

1. How This Business Actually Works

Msscorps sells reports, not parts. A customer ships a wafer fragment or a defective packaged device into a lab; technicians prepare a sample (often a lamella thinner than a virus); a high-end electron microscope or ion beam takes the chip apart at the nanometre scale; a few days later a multi-page PDF report goes back. Pricing is per case, not per wafer or per machine-hour. The annual report explicitly states pricing is "based on customer demand, not entirely on quantity," and the company books a single consolidated revenue line — Analysis of Service Revenue — for 100% of the top line.

The economic engine is therefore utilisation × billing rate, less a fixed depreciation block that ratchets up whenever a NT$200M+ tool lands on the floor. The lab that buys three TEMs in FY24 eats the depreciation immediately and earns the revenue 12–24 months later. This is why Msscorps' gross margin compressed from 40.3% in FY2022 to 26.7% in FY2024 and to roughly 20% on management's FY2025 guide — the same business with newer machines. The customer side is sticky: once a lab is qualified into a foundry's R&D flow (often with a dedicated server/network link), switching is operationally painful, which is why the top-2 customers have stayed at ~22% of revenue each and analysts cover the Taiwan trio (MA-Tek, iST, Msscorps) as a single monthly revenue series. Bargaining power leans toward the lab on technically difficult work (sub-3nm cross-sections, silicon photonics) and toward the customer on routine FA.

2. The Playing Field

Msscorps sits inside a three-firm Taiwan oligopoly that dominates independent MA/FA capacity globally. The right comparison set is MA-Tek and iST; everyone else listed in Taiwan testing-related space is in a different business model.

Three readings come out of this table. First, the analysis labs (Msscorps, MA-Tek, iST) are an order of magnitude smaller than the production-test peers; using "semiconductor testing" market-size reports for Msscorps implicitly drags in KYEC's volume-driven pool, most of which is structurally inaccessible to a TEM/FIB lab. Second, the margin gap is a capex-cycle gap, not a moat gap: MA-Tek (FY24 GM 32.9%) and Msscorps (26.7%) are sub-scale to KYEC's 34.8% only because they refresh tools faster than they grow utilisation. Third, Msscorps trades at a meaningful premium to MA-Tek on EV/Revenue (~20x vs MA-Tek's ~5x and iST's ~3.5x at the same FY24 base) — the market is pricing the FY26 recovery, not the FY24 numbers. MA-Tek remains the operational benchmark: same niche, more diversified service mix, and four years of consistently double-digit operating margin while Msscorps' OM swung from 21.9% (FY22) to 6.85% (FY24). What "good" looks like in this industry is MA-Tek's combination of 32–40% GM and 14–19% OM through both up- and down-cycles.

3. Is This Business Cyclical?

Yes — but the cycle hits margin first, revenue second, and pricing barely at all. Demand comes from foundry R&D budgets, which historically grow even when wafer-fab capex contracts; foundries cannot afford to fall behind on 2nm or HBM4 because mature-node utilisation softened. That makes the top line less cyclical than its customers. The capex side is the opposite: tool lead times of 12–18 months force labs to commit before they know utilisation, so depreciation rises ahead of revenue and falls behind it.

The depreciation-only chart looks brutal: gross margin nearly halved over three years on a revenue base that grew 14% per year. But operating cash flow held steady at NT$650–740M across FY22–FY25 because the missing gross profit is largely non-cash. The real cycle indicator is capex intensity: above ~30% of revenue signals build-out (margin compression to follow); below ~15% signals harvest (margin tailwind to follow). Msscorps just turned that corner — FY24 peaked at NT$1.67B of capex (85% of revenue); FY26 guidance is around NT$500M (~17%). Revenue, by contrast, did not collapse in any chip down-cycle in the file: FY22→FY23 grew 9% and FY23→FY24 grew 5% straight through the worst memory and PC downturn since 2009.

4. The Metrics That Actually Matter

Five numbers explain almost everything important about value creation in this business — and four of them are not on the standard income statement.

Scorecard heat — positive = improving, negative = under stress; scale -2 to +2.

The metric to internalise is capex intensity, not the P&L. Gross margin is downstream of it by a year; ROE is downstream by two. A reader who only watches reported earnings will systematically misjudge this stock — selling when depreciation is loading and buying after recovery is priced in. The contrast with standard semiconductor screens is sharp: P/E and EV/EBITDA understate value during build-out and overstate during harvest. Msscorps' depressed FY24/FY25 earnings include real, paid-for production capacity that has not yet shown up as revenue.

5. What Is This Business Worth?

Value here is normalised earnings power, not current earnings. Msscorps is a single economic engine — one revenue line, one cost structure, one geographic concentration — so sum-of-the-parts is not the right lens; trying to value the Japanese or US lab subsidiary separately would only restate the same per-report economics at a smaller scale. The right valuation question is what the lab earns once capex normalises and the FY24-vintage fleet hits target utilisation.

At NT$780 the market is paying ~50x trailing EBITDA and ~20x trailing revenue against MA-Tek's ~11x and ~4x on roughly the same business. That premium only makes sense if FY26 delivers: revenue NT$2.7–2.8B at ~30% GM would generate EBITDA roughly double the trailing figure, dropping forward EV/EBITDA toward the 20–25x range that MA-Tek and Ardentec command. Conversely, a half-delivered guidance (NT$2.4B at 24% GM) leaves the stock paying for a recovery that did not arrive. The clearest underwriting frame is therefore: buy if you believe FY26 utilisation hits guidance; the multi-year IRR is built on that single hinge.

6. What I'd Tell a Young Analyst

Three things, none of them about the headline number.

Watch capex intensity and gross margin together — never alone. A standalone GM decline in this industry usually means tool refresh, not lost pricing. Look at the lag: GM bottoms 4–6 quarters after capex peaks. Msscorps' capex peaked in FY24; the GM bottom is the FY25 print; the FY26 print is the rebound. If FY26 GM is still in the low-20s by Q2, the thesis has broken — not because the lab is bad, but because the new tools aren't busy enough.

Track the monthly revenue trio (MA-Tek, iST, Msscorps) on TWSE MOPS. It is the highest-frequency demand signal in the file. A divergence where Msscorps lags the other two by more than 5 percentage points YoY for two consecutive months is the cleanest "losing share" warning the data supports. Conversely, all three printing record monthly highs is a sector all-clear.

The single biggest catalyst is named-customer disclosure. Top-2 customers are each ~22% of revenue and the 17 Dec 2025 briefing flagged an AI-zone anchor scaling from 25% to 75%. Confirmation of a Nvidia-, Broadcom-, or hyperscaler-tier name in the top three would convert the concentration footnote from risk into a quality signal. If Customer A or B exits — the data does not say who they are — there is a 20%+ revenue gap with no quick offset. Concentration is both the bull case and the largest concentrated downside in the same number.

What the market is most likely overestimating: the linearity of the FY26 ramp. Capex of this size rarely fills neatly to schedule; even MA-Tek's much steadier business swung GM by 6pp between FY22 and FY24 on similar dynamics. What the market is most likely underestimating: the durability of method-IP. The MA-Tek litigation history is not a distraction — it is the clearest evidence that working methods, technician retention, and patent filings are the actual moat in this industry. Everything else is depreciation.