DEVICE CO., Ltd (187870.KQ) - Deep Dive Research Report

Prepared 2026-06-13. All figures in Korean won (KRW) unless stated. No valuation, price-target, or market-cap data for the subject company appears in this report.

A note on sourcing. Device is a small-cap KOSDAQ company. Like nearly all KOSDAQ small-caps, it does not host publicly available quarterly earnings conference calls with English (or even Korean) transcripts. There is no Seeking Alpha / company-hosted concall archive for this name. Where the report's template calls for "5 concalls," I have substituted the five most recent statutory quarterly reporting periods filed to Korea's DART system (Q1 2026, FY2025/Q4 2025, Q3 2025, Q2 2025, Q1 2025), cross-read against the company's contract disclosures (단일판매·공급계약), Korean trade press (THE ELEC, KIPOST, The Bell, etc.), and the MoatMap insider feed. Every forward-looking statement in Sections 7 and 9 is dated to the filing or disclosure it came from. This is disclosed rather than papered over.

SECTION 1: WHAT THE COMPANY DOES

Device makes the machines that clean the most sensitive consumable tools inside an OLED display fab and a semiconductor fab. It is not a chip company and not a panel maker. It sells capital equipment, and a slice of recurring materials, to the companies that build screens and chips.

Two products carry the business:

1. Fine Metal Mask (FMM) / shadow-mask cleaning equipment for OLED manufacturing. When an OLED panel is built, red, green and blue organic material is evaporated onto a glass (or, increasingly, a large 8.6-generation) substrate. To land each colour in exactly the right sub-pixel, the vapour passes through a fine metal mask - an Invar foil roughly 15-30 microns thick, perforated with millions of microscopic apertures. Over repeated evaporation cycles the mask clogs with organic residue. If you do not clean it precisely, the apertures distort, the colours bleed, and yield collapses. But the mask is delicate: clean it too aggressively and you warp the foil or widen the holes, which is just as fatal. Device's machines remove the organic build-up without deforming the mask, then optically inspect it. This is the company's flagship.

2. FOUP (Front-Opening Unified Pod) cleaning equipment for semiconductors. A FOUP is the sealed plastic carrier that shuttles a batch of wafers between process tools in a fab. Over time it absorbs and then outgasses airborne molecular contamination (acids, bases, moisture) and sheds particles onto the wafers it is meant to protect. At advanced nodes - especially EUV lithography - this contamination causes defects, haze and reticle problems. Device's FOUP cleaners degas, wash the pod body and door separately, and dry the carrier so it stops poisoning wafers.

Founding and evolution. The company was founded on 1 October 2002 by Choi Bong-jin (최봉진), who remains co-CEO and the controlling shareholder. It built its reputation as the exclusive mask-cleaning supplier to Samsung Display from the earliest small/mid OLED lines - the lines that went into Galaxy phones and, later, the iPhone. That anchor relationship is the spine of the whole company: being the qualified, designed-in cleaner on the world's largest small/mid OLED maker's lines is what let Device then sell the same capability into China's panel build-out (BOE, Tianma, CSOT, Visionox). The company listed on KOSDAQ on 20 December 2017, and on 1 April 2025 it shortened its name from Device ENG (디바이스이엔지) to simply Device (디바이스).

The core value proposition. Device sells yield protection for the most expensive, most contamination-sensitive consumable in two of the hardest manufacturing processes on earth. A panel maker spends a fortune on FMMs and on evaporation tools; a poorly cleaned mask wastes both. A chipmaker running EUV cannot afford a FOUP that contaminates a wafer worth thousands of dollars. Device's machines are a small line item that protects a very large one. That asymmetry is the whole pitch.

Why it is hard to replicate. The difficulty is not in "washing." It is in removing organic contamination from an ultra-thin, precision-perforated foil (or a chemically reactive plastic pod) without altering the thing you are cleaning, repeatably, at production speed, and proving it with inline optical inspection - on tools that the customer has qualified over years on its own lines. The qualification barrier, not the mechanical design, is the moat.

Device's own framing, in its IPO-era pitch: it positioned itself as the company with "domestic-unique semiconductor and display cleaning technology" that supplies "the world's largest small/mid OLED maker and semiconductor makers" with the relevant cleaning equipment. The whole business is built on being the one qualified supplier where qualification is brutally hard to win.

SECTION 2: BUSINESS SEGMENTS

Device reports primarily through one operating line ("display and semiconductor contamination-removal equipment," historically ~99%+ of sales) plus a small materials/components activity. Functionally, though, the business splits into two end-markets with very different dynamics. I treat them as segments because the customers, competitors and demand cycles diverge sharply.

2.1 Display cleaning (FMM / shadow-mask) - the historical core (~70-80% of revenue)

What it does. Supplies mask-cleaning and mask-inspection equipment to OLED panel makers, plus deposition-chamber cleaning. The end market is small/mid OLED (smartphones, smartwatches) and, now, the new wave of 8.6-generation IT OLED (tablets and laptops). Customers are Samsung Display in Korea and the Chinese panel cluster - BOE (Chongqing B12, Chengdu 8.6G), Tianma (Xiamen, Wuhan), CSOT (Wuhan), Visionox (Hefei).

Core capability. The know-how is cleaning a 15-30 micron Invar FMM without distorting its apertures, plus the inline optical inspection that proves the mask is fit to reuse. This was co-developed with Samsung Display over years on live lines. You cannot buy that qualification; you earn it one line at a time.

Why it exists as a distinct business. It is a display problem (evaporation, masks, organic chemistry), sold to panel customers on panel capex cycles - completely different from the semiconductor buyer. Its competitors are different too.

Competitive position. Device's strength is the Samsung Display anchor and a multi-year China track record. But it is not an unchallenged monopoly: in FMM cleaning it competes with Japan's Okawa in China, and Korean peers have pushed into the new 8.6G FMM cleaning opportunity - KC Tech (케이씨텍) publicly won an "industry-first 8.6G FMM cleaning" order, and STI (에스티아이) has won Samsung QD-display mask cleaning. So the historical near-monopoly on small/mid lines is being contested precisely as the market transitions to 8.6G.

Role in the group. This is both the cash engine and the cyclical growth bet. The 2025 recovery was overwhelmingly a display story.

2.2 Semiconductor cleaning (FOUP / wafer-carrier) - the diversification leg (~20-25% of revenue)

What it does. Supplies FOUP cleaning and contamination-removal equipment to chipmakers. End markets: advanced foundry/logic (EUV lines) and memory (DRAM/HBM). Customers are Samsung Electronics and SK Hynix.

Core capability. Degassing and separately cleaning the FOUP body and door to control airborne molecular contamination at advanced nodes. The marquee proof point: Device won the FOUP-cleaning slot on Samsung's Hwaseong EUV-dedicated foundry line ahead of US incumbent Brooks Automation - a genuine displacement of a foreign incumbent at the most demanding node.

Why it exists separately. Different physics (molecular contamination in a plastic pod, not organic residue on a metal foil), different customers (Samsung Electronics / SK Hynix, not Samsung Display), and a different cycle (memory/HBM and foundry capex, not panel capex). It was deliberately built as a second leg to reduce reliance on the display cycle.

Competitive position. This is a more crowded field: Brooks Automation/Azenta and Entegris (US), Gudeng (Taiwan), 3S Korea, and the fast-rising ISTE (아이에스티이), which supplies SK Hynix HBM lines. Device wins on Samsung qualification and the EUV reference; it is more exposed on the SK Hynix side, where ISTE has momentum.

Role in the group. The strategic option - smaller today, but it rides HBM and EUV, the strongest structural tailwinds in semis. New Samsung wafer-cleaning supply contracts in late 2025 and 2026 show it is live and growing.

Segment summary

SECTION 3: PRODUCTS AND BUSINESS DETAIL

Product catalogue:

• FMM / shadow-mask cleaning equipment. Removes organic deposition residue from fine metal masks between evaporation cycles so the mask can be reused without aperture distortion. Used by OLED panel makers. This is the product on which the company was built and the one driving the 2025-2026 upcycle (the BOE 8.6G order below).
• FMM optical inspection equipment. Inspects the cleaned mask to verify aperture integrity - sold alongside the cleaner as part of the mask-management workflow.
• OLED deposition-chamber / process cleaning equipment. Higher-efficiency cleaning of the deposition process environment.
• FOUP cleaning / contamination-removal equipment. Degasses and cleans wafer-carrier pods (body and door separately) for semiconductor fabs; the EUV-line variant is the most demanding.
• Materials and components for semiconductor and display processes - a smaller, more recurring activity layered on top of the equipment sales to diversify revenue.

What makes the products hard to make: the binding constraint is customer qualification, not raw mechanical engineering. A cleaning tool has to be proven, on the customer's own line, to remove contamination to spec without damaging a fragile, expensive consumable, and to do it at production throughput with inline verification. That qualification took years on Samsung Display's lines and on Samsung Electronics' EUV line, and it is what a new entrant cannot shortcut.

Manufacturing and geography. Device is headquartered and manufactures in Cheonan, South Chungcheong Province (Seongeo-eup). It sells domestically to the Samsung complex and SK Hynix, and exports heavily to the Chinese OLED cluster. China has been the swing factor: the 2019 FMM orders from Wuhan Tianma (~₩12.3bn) and Hefei Visionox (~₩48.9bn) and the 2024 BOE 8.6G order (~₩68.5bn) all came from China's panel build-out.

Milestones that changed the business:

• 2002 founding; 2017 KOSDAQ listing.
• 2019 large FMM cleaning orders from Chinese panel makers (Tianma, Visionox) - the first China inflection.
• 2020 peak year (revenue ~₩126bn) driven by FMM demand, followed by three lean years.
• ~2021-2024 displacement of Brooks Automation to win the Samsung Hwaseong EUV-line FOUP cleaning slot - the semiconductor leg's breakout proof point.
• September 2024 the ₩68.5bn BOE 8.6G FMM cleaning order (Chengdu) - the contract that drove the entire 2025 recovery.
• April 2025 rename to Device; 2026 corporate actions (bonus issue, third-party allotment, founder/management cluster buying, absorption of subsidiary OClen Ventures, and addition of solar-power generation to the articles of incorporation as a future business option).

SECTION 4: CUSTOMERS

Who buys, specifically:

• Samsung Display - the anchor display customer; Device has been its exclusive mask-cleaning supplier since the early small/mid OLED lines.
• Samsung Electronics - FOUP cleaning for advanced foundry (Hwaseong EUV) and memory/HBM. Repeated single-supply contracts in 2025-2026 (e.g. ~₩2.97bn Dec 2025-Jun 2026; ~₩6.95bn Apr 2026-Feb 2027).
• SK Hynix - FOUP cleaning for memory/HBM lines.
• Chinese OLED makers - BOE, Tianma, CSOT, Visionox - for FMM cleaning as they build out OLED capacity, now at 8.6G.

Who makes the decision and on what criteria. These are fab process-engineering and equipment-procurement organisations. The buying criteria are qualification track record, cleaning performance proven on the customer's own process, throughput, and reliability - price matters but comes after "will this tool protect my yield on this exact line." Sales cycles are long: a tool gets designed into a new line build, qualified, then ordered, with revenue recognised as lines ramp.

Why they choose Device. For Samsung Display, incumbency and a decade-plus of co-development. For Samsung Electronics' EUV line, a cleaning result good enough to displace Brooks. For Chinese panel makers, an installed reference base and a credible alternative to Japan's Okawa.

Switching costs. High on existing lines - re-qualifying a different cleaning supplier on a running line risks yield, so incumbents are sticky. But switching costs reset at each new line / each new generation: an 8.6G build is a fresh bake-off, which is exactly why KC Tech and STI can contest the next wave even though Device owns the legacy small/mid lines.

Concentration. Very high, and it cuts both ways. Revenue is concentrated in the Samsung complex plus a handful of Chinese panel makers. The 2025 surge came overwhelmingly from a single BOE order. That concentration is a quality signal (you only sell to these names if you are qualified) and a risk (any one program slipping swings the P&L hard - visible in the lumpy quarter-to-quarter revenue).

Contract structure. Lumpy, project-based single-supply contracts (단일판매·공급계약) tied to line builds, layered with a thinner recurring materials/components stream. Revenue predictability is low quarter-to-quarter and is best read on a 12-month, order-book basis rather than sequentially.

SECTION 5: COMPETITIVE LANDSCAPE

The industry structure is "specialist cleaning-equipment vendors qualified into specific fabs." There is no single dominant global cleaner; the field fragments by product (FMM vs FOUP) and by customer.

In FMM / display mask cleaning, Device's historical edge is the Samsung Display incumbency and China references. Its real contest is (a) Japan's Okawa in China, and (b) Korean peers moving in for the 8.6G transition - KC Tech, which announced an industry-first 8.6G FMM cleaning win, and STI, which won Samsung QD-display mask cleaning. The risk here is that the generational reset to 8.6G is exactly when an incumbent can be dislodged.

In FOUP / semiconductor cleaning, Device competes with Brooks Automation/Azenta and Entegris (US), Gudeng (Taiwan), 3S Korea, and ISTE (Korea). Device's signature win is displacing Brooks on Samsung's EUV line; ISTE is the rising threat on the SK Hynix HBM side.

Barriers to entry are real but local: per-line, per-customer qualification. They are high enough to keep the field small, but low enough that a credible domestic peer can win the next generation's bake-off. This is not a wide structural moat; it is a set of sticky incumbencies that must be re-defended each cycle.

Where Device is strong: Samsung Display mask cleaning, the Samsung EUV FOUP reference, and China FMM exports. Where it is exposed: the 8.6G FMM bake-offs against KC Tech/STI, and the SK Hynix FOUP business against ISTE.

Market-cap figures are rough peer-size references as of mid-2026 and move daily; the Korean small/mid-cap entries are left indicative rather than precise to avoid false precision. They are included only to convey relative scale, per the report's competitor-only market-cap allowance.

SECTION 6: INDUSTRY

What drives demand. Two distinct cycles:

1. OLED capex, now turning to 8.6-generation IT OLED. The structural story of 2026-2027 is mid-size OLED moving from trial to mainstream - tablets and laptops adopting OLED, which requires new 8.6G evaporation lines, which require new masks, which require mask cleaning. Samsung Display is bringing an 8.6G IT OLED fab (a ~₩4.1tn investment, ~15,000 sheets/month) online from Q2 2026; BOE is building an 8.6G line in Chengdu (a ~₩12tn / ~63bn RMB investment, estimated ~32,000 sheets/month at glass-substrate scale), with FMM phase-1 set and phase-2 equipment orders flagged for early 2026. This is the direct demand source for Device's flagship product.
2. Semiconductor capex - EUV foundry and HBM memory. FOUP cleaning demand rides advanced-node logic and the HBM build-out. HBM in particular has driven incremental Samsung FOUP-cleaning orders.

Industry size and trajectory. Device sits in a niche (mask-cleaning and FOUP-cleaning equipment) within the much larger OLED-equipment and semiconductor-equipment markets. The relevant driver is not the absolute market size but the 8.6G OLED transition, widely described in Korean trade press as the 2026-2027 inflection where mid-size OLED becomes a mainstream market and Korea (Samsung Display) and China (BOE) race to mass production.

Position in the supply chain. Device is a Tier-1 process-equipment supplier to panel and chip makers - upstream of the panel/chip, downstream of raw materials. Its product is a small but yield-critical node in the OLED evaporation and semiconductor wafer-handling chains.

Import-substitution dynamic. On the semiconductor side, Device is part of Korea's localisation push, displacing US Brooks on a Samsung EUV line - a "domestic equipment" substitution story. On the display side, it is the export side of that coin, competing against Japan's Okawa for Chinese demand.

Regulation / certification. No external regulatory approval gates the product; the binding "certification" is private customer qualification. The 8.6G OLED race carries geopolitical overtones (Korea vs China display leadership), but the equipment itself is not export-controlled in the way leading-edge litho is.

Cyclicality. High and lumpy. The business follows panel and chip capex, which is famously cyclical - Device printed a peak year in 2020, then three down years, then a sharp 2025 recovery. Revenue is concentrated in a few large project orders, so even within an up-cycle the quarters are uneven.

Tailwinds: 8.6G IT OLED build-out (Samsung Display + BOE), HBM and EUV capex, Korean equipment localisation. Headwinds: dependence on a small number of customer capex decisions, Chinese panel-maker financial health, and intensifying domestic competition at exactly the 8.6G transition point.

SECTION 7: GROWTH TRIGGERS

Source caveat (repeated): Device holds no public earnings calls. The triggers below are drawn from statutory quarterly filings and dated contract/press disclosures, each cited to its reporting period or announcement date. They are management/company-disclosed forward drivers, not analyst opinion.

• BOE 8.6G FMM cleaning ramp and a flagged phase-2 order. The ~₩68.5bn BOE Chengdu 8.6G FMM cleaning contract (disclosed September 2024) drove the 2025 revenue surge, and trade press reports a second BOE investment phase of similar scale with additional 8.6G equipment orders flagged for early 2026.

  Per The Bell's FY2025 review (Feb 2026): "BOE is reportedly launching a second investment phase matching the initial project's scale, positioning Device for continued growth."

• Samsung Display 8.6G IT OLED line start-up (from Q2 2026). Samsung Display's ~₩4.1tn 8.6G IT OLED fab comes online from Q2 2026; as the incumbent Samsung Display mask-cleaning supplier, Device is positioned for mask-cleaning content on that line. (Industry disclosure, reported H1 2026.)
• New Samsung Electronics FOUP / wafer-cleaning contracts. A ~₩2.97bn wafer-cleaning single-supply contract (disclosed 8 Dec 2025, running to 30 Jun 2026) and a larger ~₩6.95bn contract (disclosed for 9 Apr 2026 - 26 Feb 2027), tied to Samsung memory/foundry capex. (Contract disclosures, Dec 2025 and 2026.)
• HBM-driven semiconductor cleaning demand. Management/press attribute incremental FOUP-cleaning orders to Samsung's HBM build-out; ~₩16.9bn of new Samsung semiconductor contracts were flagged in late 2024 on HBM demand. (FY2024-2025 disclosures.)
• IT-OLED-led Q1 2026 momentum. The Q1 2026 reporting period showed revenue up ~260% year-on-year, attributed to expanded IT OLED investment ("IT OLED 투자 확대로 실적 개선"). (Q1 2026 quarterly filing.)
• Materials/components and adjacency expansion. Continued push into process materials/components, plus the March 2026 addition of solar-power generation to the articles of incorporation as a future business option, and absorption of subsidiary OClen Ventures (effective 31 Jul 2026) to streamline operations. (AGM 31 Mar 2026; merger disclosure 2026.)

SECTION 8: KEY RISKS

• Single-program, single-customer concentration. The 2025 surge rode largely one BOE order and the Samsung complex. Mechanism: if a major OLED program slips, is cancelled, or a Chinese customer defers capex, a large share of revenue evaporates in a quarter. The lumpy quarterly P&L (Q1 2025 operating loss, Q3 2025 huge profit) shows how violently a single program timing swings results. Calibration: high-probability moderate-to-severe drag - it is the defining feature of the business model, not a tail event.

• 8.6G competitive reset. Mechanism: generational transitions re-open the bake-off. Device's incumbency is on legacy small/mid lines; KC Tech publicly won an industry-first 8.6G FMM cleaning order and STI won Samsung QD mask cleaning. If Device loses 8.6G FMM share to domestic peers, its core franchise erodes precisely as the market it depends on grows. Calibration: medium-probability, high-impact - this is the most important strategic risk.

• Chinese demand and customer financial health. Mechanism: a large slice of FMM demand is Chinese panel makers (BOE, Tianma, CSOT, Visionox). Their capex is policy- and subsidy-driven and can stop abruptly; export receivables to Chinese customers carry collection and currency risk. Calibration: medium-probability moderate drag, correlated with China's display-industry cycle.

• SK Hynix FOUP share loss to ISTE. Mechanism: ISTE is a credible, fast-rising FOUP-cleaner supplier into SK Hynix HBM lines. If Device fails to win meaningful SK Hynix content, its semiconductor diversification leans almost entirely on Samsung. Calibration: medium-probability, moderate impact on the smaller segment.

• Capital-structure dilution. Mechanism: in 2026 the company executed a bonus issue (6,733,451 new shares) and a third-party allotment paid capital increase (9,212,000 new shares). The bonus issue is cosmetic, but the allotment materially raises the share count. If the raised capital is not deployed into returns-accretive growth, per-share value is diluted. Calibration: certain event, impact depends on capital deployment.

• Cyclical whipsaw / over-extrapolation. Mechanism: the company has a documented history of one peak year (2020) followed by three lean years. A reader anchoring on the 2025-2026 upcycle risks over-extrapolating; when the 8.6G order wave is filled, revenue can fall as sharply as it rose. Calibration: high-probability over a multi-year horizon - this is a capex-cycle equipment vendor, not a compounder.

• Diversification dilution of focus. Mechanism: adding solar-power generation to the charter and pursuing materials/components could spread management attention and capital away from the cleaning franchise. Calibration: low-probability material harm today, worth monitoring.

SECTION 9: WALK THE TALK

Reporting periods used (no concalls exist for this name): Q1 2025, Q2 2025, Q3 2025, FY2025/Q4 2025 (results published Feb 2026), and Q1 2026 (most recent, filed ~May 2026 - within 90 days of today). Because there are no transcript "guidances," I assess credibility against what the company disclosed it had contracted versus what subsequently showed up in the numbers.

The cleanest "promise vs outcome" thread is the BOE 8.6G order. In September 2024 the company disclosed the ~₩68.5bn BOE FMM cleaning contract and signalled revenue would flow from late 2024. That is testable. It delivered: FY2024 revenue was ~₩47.2bn (already recovering off the trough), and FY2025 revenue reached ~₩84.1bn (+78%) with operating profit ~₩17.3bn (+311%) - "three years of decline ended," driven exactly by the BOE display orders management had pointed to. The order converted into the revenue and margin recovery the company had implied. That is a kept promise, and the most important one.

The quarter-by-quarter pattern corroborates a management that under-promises on smoothness and lets lumpy programs land when they land. Q1 2025 was an operating loss (~-₩0.8bn) on ~₩8.6bn revenue; Q2 modestly profitable; then Q3 2025 exploded to ~₩39.2bn revenue and ~₩13.3bn operating profit as BOE shipments concentrated; Q4 normalised to ~₩27.0bn revenue. Crucially, the recovery did not fizzle into 2026: Q1 2026 printed ~₩31.1bn revenue (+~260% YoY) and ~₩12.3bn operating profit, attributed to expanded IT OLED investment. So the "continued growth from 8.6G / IT OLED" thesis the company carried into 2026 is, so far, being delivered rather than slipping.

The FOUP/semiconductor diversification claim also walks: the company said it was winning Samsung semiconductor cleaning content, and the contract record backs it - the EUV-line displacement of Brooks, plus dated single-supply contracts (~₩2.97bn in Dec 2025, ~₩6.95bn in Apr 2026). These are small relative to the display surge, but they are real and disclosed, not vapour.

Where I would flag caution rather than broken promises: the 2026 capital actions. The bonus issue plus a 9.2-million-share third-party allotment are shareholder-friendly in optics (bonus issue) but dilutive in substance (allotment), and the company has added businesses (solar charter language, OClen absorption) during an up-cycle. None of this is a missed commitment, but it is the classic pattern of a cyclical equipment vendor raising and reshaping capital at the top of a wave - worth holding management to account on deployment.

Assessment: on the operational promises that matter - converting the BOE order into a revenue/margin recovery and carrying it into 2026 - this management has done what it said. The honest qualification is that the "promises" here are contract conversions, not concall guidance, and the 2026 capital-raising-plus-diversification pattern is the kind of top-of-cycle behaviour that warrants follow-through scrutiny. Net read: credible operationally, watch the capital allocation.

SECTION 10: SHAREHOLDER FRIENDLINESS INDEX

Dividends. Device pays a small but sharply growing ordinary dividend: DPS of ₩51 (FY2023), ₩102 (FY2024), and ₩332 (FY2025) - roughly a doubling then a tripling, tracking the earnings recovery (source: company/Company-Guide dividend records). The trend is unambiguously up and pro-cyclical: the company shares the upcycle with holders rather than hoarding all of it, though the absolute payout remains modest relative to the swing in earnings.

Buybacks and dilution. This is mixed and must be split by window. Recent (last ~90 days, MoatMap): no share buybacks recorded since 15 March 2026 - an empty buyback window, not evidence of a multi-year policy. Older / three-year context (external sources): the company holds treasury shares from prior repurchases and in December 2025 disposed of 101,420 treasury shares (~₩1.4bn) to fund employee/executive performance bonuses - i.e. it was releasing treasury stock, not retiring shares. On dilution, the direction in 2026 is clearly toward more shares, not fewer: a bonus issue of 6,733,451 shares (record date 29 Apr 2026) and a third-party allotment of 9,212,000 new shares (record date 9 Jun 2026) sit on top of a ~13.77 million prior share count - so shares outstanding are rising materially, and the treasury disposal adds a touch more float. I could not verify a formal multi-year buyback-and-cancel program; if one existed it was not material, and the net three-year trajectory is dilutive, partly offset by a rising cash dividend.

Verdict: Neutral, tilting shareholder-friendly on dividends but dilutive on share count - the rapidly growing DPS rewards holders, but a bonus issue plus a sizeable capital-raising allotment in 2026 mean the share count is expanding, so per-share generosity is being diluted by issuance at the same time.

SECTION 11: INSIDER ACTIVITIES

Source: MoatMap cross-market disclosure database (market: KR), the canonical source for recent Korean DART insider filings, current to 2026-06-12 23:00 UTC. Korea's DART portal is API-gated and not searchable via open web, so per the venue rule this block is the sole source for recent insider dealing.

The story here is emphatic: a broad-based insider buying cluster in May-June 2026, led by the founder.

Reading the buys. Every reportable insider transaction in the window is a purchase at ₩23,950 on 22 May 2026, spanning the founder, the co-CEO, and at least eight other executives and directors. This is textbook cluster buying - the single most bullish insider configuration there is, because it is hard to explain except by shared conviction. The founder Choi Bong-jin's purchase is enormous in context: tens of billions of won, lifting his ownership by mid-teens-percent of the company. A purchase of this scale by the founder-CEO, alongside the co-CEO and the entire senior bench buying in the same session, is a very bullish signal. The likely mechanism is the founder (and management) subscribing to the company's 2026 third-party allotment and adding in the market - i.e. management putting fresh capital into the company at the same price, not merely reshuffling existing holdings. The subsequent Mirae Asset 5%-rule buy on 2 June adds an institutional vote of confidence on top.

Sells. There were no insider sells in the trailing twelve months in the data. The only non-buy line is a 7 May 2026 zero-share "Other" entry for the founder - a deemed-interest/5%-rule housekeeping report with no directional content, not a disposal.

Net assessment. Insiders are overwhelmingly net buyers, and the buying is broad-based rather than concentrated in one person - founder, co-CEO, multiple EVPs/directors, and an institution all bought within a two-week window, with zero offsetting sales. This coincides with the operational recovery and the 8.6G order wave. Plain-language read: bullish signal - management is backing the up-cycle with its own money. The one sober caveat (Section 10) is that some of this buying is the founder funding a capital raise that also dilutes outside holders; the conviction is real, but it arrives bundled with issuance.

SECTION 12: SCENARIOS

Bull case. The 8.6G IT OLED transition turns out to be the multi-year super-cycle the trade press expects. Samsung Display's new 8.6G fab ramps from Q2 2026 and Device, as the incumbent mask-cleaning supplier, lands the cleaning content. BOE follows its first Chengdu phase with the flagged second phase, and Device wins it against Okawa, repeating the ₩68bn-class order. On the semiconductor side, HBM and EUV keep pulling FOUP-cleaning orders from Samsung, and Device finally cracks meaningful SK Hynix content too, blunting the ISTE threat. The founder's huge 2026 share purchase looks prescient: the capital raised funds capacity for the 8.6G wave, and the lumpy business strings together back-to-back up years instead of reverting. The materials/components and even solar adjacencies start to smooth the cyclicality. In this world, Device is the cleaning arms-dealer to the entire mid-size OLED build-out across Korea and China.

Base case. Device delivers roughly what the order book implies. The 2025-2026 recovery is real and continues through the current 8.6G order wave, with Q1 2026's strength carrying for several more quarters as BOE and Samsung Display lines ramp. But the business stays lumpy and concentrated: a couple of huge quarters, a couple of soft ones, revenue best read year-over-year. Device holds its Samsung Display mask-cleaning incumbency and its Samsung EUV FOUP slot, splits the 8.6G FMM opportunity with KC Tech/STI rather than owning it outright, and keeps semiconductor as a smaller second leg. Dividends keep rising with earnings; the 2026 share issuance dilutes per-share figures somewhat. A solid up-cycle for a cyclical equipment vendor - neither a re-rating story nor a disappointment.

Bear case. The 8.6G order wave proves to be a pull-forward, not a plateau. Once the BOE Chengdu and Samsung Display lines are equipped, new orders thin out, and - as in 2021-2023 after the 2020 peak - revenue falls hard. Worse, Device loses the 8.6G FMM bake-offs to KC Tech and STI, so it does not even capture full share of the wave it depended on, while Chinese panel customers, squeezed financially, defer the second-phase capex. On the semiconductor side ISTE takes the SK Hynix FOUP business and Device's diversification stalls at "Samsung-only." The 2026 capital raise then looks ill-timed: more shares outstanding against shrinking earnings, with the solar-power detour absorbing cash and attention. The founder's conviction buy cushions sentiment but cannot offset a capex cycle that has rolled over. Device reverts to what it was from 2021-2024 - a thin-margin, single-customer-dependent cleaner waiting for the next order.

A few honest limitations on this report: Device holds no public earnings calls, so Sections 7 and 9 use dated statutory filings and contract disclosures in place of concall transcripts (disclosed at the top). Segment revenue mix is indicative from historical commentary (display ~70-80%), not an audited split. Competitor market caps in Section 5 are rough mid-2026 peer-size references. The 2026 third-party allotment's exact subscriber list was not fully confirmable from public sources, though the founder/management cluster buy at ₩23,950 strongly implies management subscription.

Sources: The Bell FY2025 review, THE ELEC - Samsung EUV FOUP supply, Korean Wikipedia - Device ENG, OLED-Info - Device Eng, Company Guide snapshot, THE ELEC - 8.6G OLED race, DigitalToday - OClen merger, Edaily - treasury share disposal, DigitalToday - bonus issue, Hankyung - Samsung wafer-cleaning order, MoatMap KR insider feed (DART, to 2026-06-12).