HANMI Semiconductor: Q2 Operating Margin Hit a Record 51.9%, but 95.5 Times Earnings Prices a Single-Generation Moat

This review resolves the prominent first-quarter data conflict. The correct consolidated K-IFRS figures for the three months ended March 31, 2026 are revenue of KRW 50.902 billion and operating profit of KRW 8.456 billion, down 65.5% and 87.9% year on year. The widely circulated KRW 147.3–147.4 billion revenue and KRW 69.6 billion operating-profit figures describe the first quarter of 2025, when revenue and operating profit increased about 90.7% and 142.5%. The discrepancy arose from mixing years, rather than from separate-versus-consolidated accounting. It is a useful warning about relying on scraped consensus pages and undated Korean press excerpts for this company.

The order-flow dates in the research brief also need correction. The KRW 150 billion SK hynix contract that took cumulative disclosed HBM TC-bonder orders to approximately KRW 358.7 billion was announced in June 2024, not June 2026. The roughly KRW 42.8 billion Griffin order, often estimated at about 15 tools, was announced in May 2025, not May 2026. In 2026, HANMI disclosed a roughly KRW 9.65 billion SK hynix contract in January and a KRW 44.2 billion HBM4 TC Bonder 4.5 Griffin contract in June, giving KRW 53.85 billion of identified 2026 SK hynix orders from those two announcements. The 2024 cumulative number must not be added again to 2026 orders.

Customer concentration remains the structural weakness. The audited 2025 filing identifies two anonymous customers responsible for 43.96% and 14.82% of annual sales, a combined 58.78%. In 2024, the two disclosed major customers accounted for 73.47%. The reduction is encouraging, but the filing does not name the accounts, so it is not possible to prove from DART alone which percentages belong to SK hynix, Micron or another buyer. Industry reporting identifies SK hynix and Micron as major HBM-equipment customers, and Micron has publicly recognized HANMI as a supplier, but exact customer attribution would be an inference.

The moat is real but narrower than the share-price narrative implies. HANMI has decades of precision-mechatronics experience, vertically integrates design, manufacturing, assembly and testing, and has accumulated process knowledge from production deployments. A memory producer will not replace a qualified bonder casually because a tool change can require recipe redevelopment, yield validation, operator training and service support. Yet SK hynix has already widened its vendor base from HANMI toward Hanwha Semitech and ASMPT. Reports that ASMPT supplied a substantial portion of SK hynix’s newer HBM4 TC-bonder purchases show that design-in creates switching costs but does not guarantee exclusivity.

The successor-technology question is unresolved. HANMI’s Wide TC Bonder is designed for wider dies and can include fluxless bonding, extending thermal compression into future HBM generations. Its original public launch target was the second half of 2026, while a later COMPUTEX description indicated early 2027. HANMI is also investing KRW 100 billion in a hybrid-bonder facility and has indicated a second-generation prototype by the end of 2026, with commercial sales expected later. Hanwha Semitech has already unveiled a second-generation hybrid bonder for SK hynix qualification, while Besi and Applied Materials have supplied development systems and ASMPT sells both thermal-compression and hybrid-bonding platforms. HANMI leads an incumbent process but has not publicly established a production position of record in the successor process.

Vertically, HANMI has proven that it can move from localization to a globally relevant process tool. Horizontally, it remains the most focused listed exposure to HBM TC bonding, but ASMPT offers a broader bonding portfolio, Besi is further advanced in direct hybrid bonding, and Hanwha has industrial capital plus access to SK hynix qualification. The company’s qualitative portrait is therefore a company in transition: an exceptionally profitable incumbent attempting to carry its customer relationships and precision-bonding expertise across a change in interconnect technology.

The share price already reflects much of the possibility. At KRW 214,500, market capitalization is about KRW 20.44 trillion, approximately 95.5 times audited 2025 net income and about 133 times 2025 free cash flow. A secondary consensus aggregation dated July 2026 put 2026 revenue near KRW 955 billion, EPS around KRW 4,144 and the consensus target price near KRW 158,778. Those figures imply about 51.8 times forecast earnings at the July 31 close, and the market price stands roughly 35% above that aggregated target. Consensus can be stale and sparse for Korean small- and mid-cap equipment companies, but the gap illustrates how heavily the market capitalizes earnings beyond the next reported year.

Vertical history and financial review

HANMI was founded in 1980 by Kwak Noh-kwon, who had worked for approximately 14 years at Motorola Korea. The original company, Hanmi Mold, was created when Korea’s semiconductor assembly industry depended heavily on imported precision molds and packaging equipment. The initial problem was industrial localization: producing tooling precise enough for semiconductor packages while reducing foreign-equipment dependence, delivery time and service friction. This starting point shaped the company’s later strategy. It learned to combine mechanical design, vision systems, thermal control and factory support rather than becoming a wafer-fabrication-process company.

The business evolved through five economically distinct stages.

The first stage, from 1980 through the 1990s, was import substitution and process credibility. HANMI’s competitive advantage was proximity to Korean packaging customers and the founder’s manufacturing background. The addressable market was modest, but solving customer-specific production problems built the engineering culture and service relationships needed for later automated equipment.

The second stage began with the 1998 introduction of Vision Placement equipment. The platform integrated package cutting, cleaning, drying, inspection, sorting and loading. It became a larger and more exportable system than the original mold business, and company materials state that it achieved the largest global market share in its category from around 2004–2005. HANMI’s lasting capability from this period was not a particular blade or camera. It was the ability to integrate multiple precision steps into a production cell that customers would run at scale.

HANMI absorbed Hanseong Electronics in 1996, adopted the Hanmi name and became HANMI Semiconductor in 2002. It listed on the KOSPI on July 22, 2005 at an IPO price of KRW 8,900, raising roughly KRW 39.6 billion. Contemporary financial data cited around the offering showed 2004 revenue of approximately KRW 69.6 billion, operating profit of KRW 19.9 billion and net income of KRW 14.8 billion. Capital markets initially understood the company as a profitable semiconductor-assembly automation exporter rather than an HBM pure play.

The third stage, spanning roughly 2010–2022, added more advanced bonding. HANMI introduced flip-chip, multi-die and large-die bonders around 2010 and developed TC-bonding equipment during the following decade. A TC-bonder-related product received Korea’s IR52 Jang Young-sil technology award in 2017. Revenue reached KRW 373.2 billion in 2021 before declining to KRW 327.6 billion in 2022. Operating margin nevertheless remained above 30% in both years, showing that the product portfolio already carried meaningful proprietary content before the HBM rerating.

The fourth stage was the 2023–2025 HBM transformation. Revenue fell 51.5% to KRW 159.0 billion in 2023 as conventional semiconductor-equipment demand weakened, but the market began to recognize TC bonding as a bottleneck in HBM production. Revenue then increased 251.5% to KRW 558.9 billion in 2024. Gross margin rose above 55%, and operating profit reached KRW 255.4 billion. The 2025 revenue record of KRW 576.7 billion was only 3.2% above 2024, while operating profit declined 1.6% to KRW 251.4 billion. The business had moved from explosive initial tool deployment into a more uneven period of customer qualification and delivery scheduling.

The fifth stage began in late 2025 and is still unfolding: supplier diversification and technology transition. HBM4 delivery timing moved between quarters, SK hynix qualified competing TC bonders, and hybrid-bonding development moved closer to production trials. HANMI responded by expanding Griffin models, introducing inspection and large-die tools, proposing a Wide TC Bonder for future HBM, and funding a dedicated hybrid-bonder plant. This stage will determine whether the company remains a category leader after the process architecture changes, or whether its peak economics belonged to the first HBM TC-bonder ramp.

† The audited filing reports KRW 332.1 billion; some data services classify expenses differently and show about KRW 330.1 billion. ‡ Rounded from historical filings and financial databases; 2024–2025 figures are from the latest audited comparative statements. Sources: HANMI’s audited filing and S&P Global/Dow Jones datasets.

The table shows why a smooth compound-growth model is inappropriate. Revenue was lower in 2023 than in any of the preceding displayed years, then more than tripled in 2024. Fixed engineering, research and factory costs allowed operating profit to rise much faster than revenue in the upturn, but the same structure produced the first-quarter 2026 collapse when customer acceptance slipped. The long-term capability is high incremental profitability, paired with low quarterly visibility.

Net income has been less informative than operating profit. The KRW 267.2 billion of 2023 net income was generated despite only KRW 34.7 billion of operating profit because investment disposals and fair-value movements produced a very large non-operating gain. The 2025 cash-flow reconciliation also included gains on financial assets and investments, although the operating business itself was highly profitable. Valuation should normalize non-operating securities gains rather than capitalizing reported net income mechanically.

Across 2021–2025, aggregate operating cash flow was about KRW 576.8 billion against aggregate net income of KRW 830.5 billion, a cash-conversion ratio of approximately 0.69. The principal reason was the 2023 investment-gain distortion, followed by working-capital movements during the HBM ramp. Annual conversion recovered to 0.93 in 2024 and 1.07 in 2025. The latest year benefited from an KRW 85.4 billion reduction in receivables but absorbed KRW 26.0 billion in additional inventory, so even the strong 2025 cash result should not be treated as a fixed run rate.

At the end of 2025, cash stood at KRW 276.2 billion, receivables at KRW 62.4 billion, inventory at KRW 150.4 billion and property, plant and equipment at KRW 238.1 billion. Total assets were KRW 813.3 billion, liabilities KRW 123.0 billion and equity KRW 690.3 billion. The absence of meaningful financial leverage sharply reduces insolvency risk. Inventory represented about 26% of annual revenue, however, and the filing recorded a KRW 28.9 billion inventory-valuation loss adjustment, up from KRW 9.7 billion. That is a significant reminder that specialized tools and components can lose value if customer specifications or delivery schedules change.

Capital expenditure increased from KRW 9.1 billion in 2022 to KRW 74.6 billion in 2025, while intangible-asset purchases added another KRW 19.1 billion in 2025. Depreciation and amortization were only KRW 11.6 billion. The gap suggests that most recent expenditure was growth investment: capacity, development facilities and new-product infrastructure, rather than maintenance of the existing factory base. A reasonable owner-earnings calculation deducts maintenance capex around current depreciation, but it must also recognize that growth investment is economically necessary if HANMI intends to compete in hybrid bonding. Treating every won above depreciation as optional would overstate distributable cash.

Capital allocation has mixed evidence. The company paid KRW 40.5 billion of dividends and repurchased KRW 189.9 billion of stock in 2024, followed by KRW 68.3 billion of dividends and KRW 7.8 billion of repurchases in 2025. It also cancelled treasury shares with an accounting value of KRW 130.3 billion in 2025. Returning capital while maintaining net cash was shareholder-friendly, but buying heavily during a rapid HBM-driven rerating exposes shareholders to timing risk. The more important allocation decision is now the KRW 100 billion hybrid-bonder investment, because unsuccessful qualification would leave a specialized facility without a corresponding profit pool. In June 2026 the company also acquired 215,600 SpaceX shares for KRW 50.0 billion, about 7.24% of year-end equity, for a 0.002% stake, citing satellite, aerospace and AI demand. That purchase has no operating link to bonding equipment, and it enlarges the securities portfolio whose fair-value movements already distort reported net income.

Management’s public targets require scrutiny. In July 2024, HANMI indicated revenue ambitions of KRW 650 billion for 2024, KRW 1.2 trillion for 2025 and KRW 2.0 trillion for 2026. Actual revenue was KRW 558.9 billion in 2024 and KRW 576.7 billion in 2025. The targets captured the direction of HBM demand but materially overstated the pace of customer orders and revenue recognition. With first-half 2026 revenue of approximately KRW 302.0 billion, reaching the old KRW 2 trillion objective would require KRW 1.698 trillion in the second half, more than three times the revenue produced by simply repeating the record second quarter twice. The historical target should be treated as ambition, not guidance. The pattern repeated more recently. At a July 30, 2025 briefing the company guided to 2025 revenue of KRW 800 billion to KRW 1.1 trillion and said it expected to win every TC bonder its principal customer would purchase for HBM4, citing roughly 90% share in HBM3E. Realized 2025 revenue was KRW 576.7 billion, and SK hynix subsequently placed HBM4 bonder orders with both ASMPT and Hanwha Semitech.

The share-price history mirrors these stages. Before HBM, investors treated HANMI as a cyclical back-end equipment exporter. The 2023–2024 rerating changed the reference point from conventional assembly machinery to an AI-infrastructure bottleneck. In 2026, the shares reached KRW 426,000 before retreating to KRW 214,500 on July 31, a decline of approximately 49.6% from the 52-week high but still 163.5% above the 52-week low. The July 31 session itself rose 27.98%, from KRW 167,600 to KRW 214,500. Such moves are too large to be explained by a one-day change in discounted equipment cash flow; they show that positioning, AI sentiment and expectations for future purchase orders dominate short-run price formation.

The valuation center shifted because both business mix and market preference changed. Operating margin above 40%, net cash and exposure to HBM justify a premium to HANMI’s pre-HBM history. A trailing multiple near 95 times audited earnings and more than 130 times audited free cash flow requires more than a premium. It requires sustained customer spending, continuing share, high margins and successful navigation of hybrid bonding. The stock is pricing a multi-year earnings path rather than the last audited year.

Business model, moat, industry, and competitors

HANMI reports one operating segment, but its economic portfolio has several layers. TC bonders are the principal narrative and likely the largest source of incremental profit. Micro SAW and Vision Placement systems remain established platforms for package cutting, cleaning, inspection, sorting and loading. Other products include HBM six-side inspection systems, bond-on-chip and chip-on-board tools for products such as GDDR and enterprise solid-state drives, large-die flip-chip bonders for 2.5D packaging, and electromagnetic-interference shielding equipment. In June 2026 it added the 2.5D TC Bonder 40, aimed at the chip-on-wafer step of CoWoS-style 2.5D packaging and handling dies from three by three millimetres up to forty by forty millimetres, alongside an FC Bonder 3.5. That line is directed at foundry and outsourced assembly customers packaging logic and accelerators, a different buyer from the memory makers who dominate current revenue. The company does not disclose product-level revenue or margin, preventing investors from separating recurring legacy-tool economics from HBM-specific economics.

In 2025, raw materials consumed about KRW 153.2 billion, employee benefits KRW 63.7 billion, outside processing KRW 28.2 billion, fees KRW 23.9 billion, depreciation and amortization KRW 11.6 billion, after-sales service KRW 11.9 billion and share-based compensation KRW 8.8 billion. Raw materials vary with shipments, while engineering employees, research, factory infrastructure and customer-service capability are difficult to reduce between orders. This cost structure produced a 51.9% operating margin in the record second quarter of 2026 and only 16.6% in the preceding quarter.

The first genuine moat is production-qualified process knowledge. TC bonding requires repeatable handling of thinner dies, accurate alignment, controlled force and temperature, limited warpage and acceptable throughput. Process recipes are developed with the customer and tested against yield targets. Once a tool enters volume production, the installed base, service history and accumulated failure data make the incumbent less risky than an unproven vendor.

A second moat comes from system integration. HANMI’s history in vision placement, cutting, inspection and automation gives it competence across mechanical handling, optics, software and process control. The company designs, manufactures, assembles and tests equipment internally. Its ability to combine several functions into an industrial tool matters more than possession of any isolated component.

The third is customer-development speed. HANMI’s focused organization can adapt a Griffin generation to a memory customer faster than a diversified conglomerate whose bonding business competes for capital with many divisions. That focus helped it become the principal supplier for early SK hynix HBM3E production. It also creates fragility: the same narrowness leaves fewer businesses to absorb a lost qualification.

The moat has clear limits. SK hynix has qualified Hanwha equipment and reportedly allocated a meaningful share of newer HBM4 tools to ASMPT. Samsung has historically used affiliate SEMES and Japan-linked alternatives, while Micron can dual-source. Equipment switching costs preserve the incumbent’s installed base, but large memory manufacturers have strong bargaining power and strategic reasons to prevent a single supplier from controlling a bottleneck. HANMI’s moat is therefore medium-to-strong within an existing customer recipe and only medium across a new HBM generation. The comparison is now quantified. Korean trade reporting puts SK hynix’s 2025 TC-bonder contract value at approximately KRW 55.2 billion to HANMI against KRW 80.5 billion to Hanwha Semitech, after SK hynix had used HANMI tools exclusively for HBM3E twelve-high production. Within a single generation the incumbent moved from sole supplier to the smaller of two.

Customer concentration quantifies this vulnerability. The two largest customers supplied 58.78% of 2025 sales, down from 73.47% in 2024. The top customer alone represented 43.96%. A single customer postponing acceptance can remove hundreds of billions of won from a year’s revenue and a disproportionate amount of operating profit. The reduction in the top-two percentage is positive only if it reflects durable new accounts rather than temporary shipment timing.

The industry is simultaneously a semiconductor cycle, capital-expenditure cycle and technology-iteration cycle. Memory producers invest when expected HBM demand, pricing and accelerator roadmaps justify capacity. Tool vendors receive orders before capacity starts, recognize revenue after delivery and acceptance, and then face a pause while the customer digests installed equipment. Technology adds another rhythm: each generation may require more alignment accuracy, thinner dies, different interconnect pitch and new material handling. HANMI benefits when both spending and process intensity rise, but a tool generation can peak before end-market HBM demand peaks.

The profit pool in advanced packaging sits where yield loss is expensive. Memory dies have already absorbed wafer-fabrication cost before stacking. A bonding defect can impair multiple dies and the base die or interposer. Customers will therefore pay for reliability and throughput, which supports unusually high equipment gross margins. Their countervailing power is qualification of multiple suppliers and internal equipment affiliates.

ASMPT estimates that the thermal-compression-bonding total addressable market can exceed US$1.6 billion by 2028, approximately KRW 2.4 trillion using a rounded KRW 1,500 per US dollar rate from July 14, 2026 reporting. TechInsights projected a 13% compound annual growth rate for HBM TC-bonder demand from 2025 to 2030. These are vendor and industry-research estimates, not audited market data, and both may include product scopes broader than HANMI’s current HBM tools.

Hybrid bonding is the largest technological uncertainty. Conventional TC bonding generally joins dies through micro-bumps under heat and pressure. Hybrid bonding directly joins dielectric and copper interfaces at finer pitch, potentially reducing stack height and improving electrical and thermal characteristics. Its challenges include surface preparation, cleanliness, wafer or die planarity, alignment, throughput and yield. Those requirements can delay adoption even when laboratory performance is attractive.

HANMI’s Wide TC Bonder is an attempt to lengthen the incumbent curve. Wider DRAM dies can increase interface area and bandwidth without relying solely on ever-higher stacks, and fluxless options may reduce contamination and package thickness. HANMI explicitly positioned the product as useful while hybrid bonding remains delayed. This is a rational response, but the two roadmap legs compete economically. Every generation that remains on TC bonding extends Griffin and Wide TC revenue; every generation that shifts to hybrid bonding makes qualification of HANMI’s hybrid platform more urgent.

The roadmap contains timing ambiguity. Initial SEMICON Korea reporting targeted the Wide TC Bonder for the second half of 2026, while a later COMPUTEX description referred to early 2027. HANMI has said it intends to launch a second-generation hybrid bonder by the end of 2026 and is building a KRW 100 billion plant in Incheon, with sales expected from 2027. A launch is not equivalent to customer qualification, and qualification is not equivalent to volume production. Investors should count a hybrid tool only when a named or credibly identifiable Tier-one customer accepts it for a production line.

The competitive landscape fits a small number of direct competitors plus several substitutes. ASMPT is the most important head-to-head listed comparison. For hybrid bonding, the strongest listed reference is Besi. A direct Korean challenger, Hanwha Semitech, sits inside the much larger Hanwha Aerospace group. Samsung’s SEMES matters as a captive competitor but is not independently listed.

† Converted at KRW 183.98 per HKD, the July 31, 2026 close. ‡ Converted at the ECB’s July 31, 2026 reference rate of KRW 1,657.99 per euro. § Hanwha Aerospace’s figures are dominated by defense and aerospace; its valuation is not a clean semiconductor-equipment comparable. Sources: company results and market data.

ASMPT became a broad packaging platform. It serves mainstream assembly, advanced logic, memory, photonics and surface-mount applications. In the first half of 2026, continuing-operations revenue reached HKD 8.90 billion, up 42.5% year on year, while bookings reached HKD 12.75 billion, up 85.1%; book-to-bill was 1.43. It offers both TC and hybrid bonding, allowing customers to remain within its platform as process architecture changes. Its weakness relative to HANMI is dilution: HBM TC bonding is one line within a much larger organization, and its reported gross margins are lower than HANMI’s peak operating margins. Its strength is resilience, global service scale and a broader set of customer qualifications.

Besi became the purest listed beneficiary of direct hybrid bonding. Second-quarter 2026 revenue was EUR 249.9 million and net income EUR 89.0 million; orders reached EUR 292.9 million, more than double the prior-year quarter. Gross margin was 65.7%, and the number of hybrid-bonding customers reportedly rose from 15 at the end of 2025 to 21 by the second quarter of 2026 across logic, memory, photonics and consumer applications. Besi’s risk is the reverse of HANMI’s: delays in hybrid adoption extend TC bonding and defer its anticipated volume. Its valuation near 71 times trailing earnings also shows that investors have already capitalized substantial adoption.

Hanwha Semitech became the best-placed Korean challenger by combining bonding development with a large industrial parent and access to SK hynix testing. It developed its first hybrid bonder in 2022, unveiled the second-generation SHB2 Nano in February 2026 and moved toward SK hynix qualification. Hanwha’s parent can fund development through a long qualification period. Its weakness is that semiconductor equipment is economically small relative to Hanwha Aerospace’s defense businesses, making public financial comparison difficult. The existence of a well-funded domestic alternative nevertheless weakens the assumption that HANMI can maintain historical HBM3E share.

The reported patent dispute between HANMI and Hanwha adds both legal and competitive risk. HANMI initiated litigation in December 2024 alleging infringement related to TC-bonder technology. Hanwha replied with invalidity petitions in May 2025 and then, in October 2025, filed its own infringement action asserting three patents covering flux application and inspection against HANMI’s bonders, with the first hearing held in April 2026. Litigation can delay products or produce damages, but it can also indicate that competitors have approached the incumbent’s commercially valuable design space. Until a final ruling or settlement appears, the dispute should not be valued as either a guaranteed defense or an impending loss.

Samsung’s equipment sourcing is strategically important even without direct HANMI revenue. Samsung has used SEMES and other tools, and reports indicate that it has tested Besi hybrid-bonding equipment. A HANMI production qualification at Samsung would weaken the single-customer bear case and validate portability of its process technology. Continued exclusion would suggest that HANMI’s moat is closely tied to SK hynix recipes rather than globally transferable.

Micron is the more visible diversification route. Industry reporting links HANMI to sizeable TC-bonder deliveries, including an order estimated at 50 tools in 2025, and Micron recognized HANMI as a top supplier. Micron is investing in additional HBM packaging capacity, including a Singapore facility expected to begin production in 2027. Micron expansion can offset SK hynix sourcing diversification, but shipment-level evidence must appear in HANMI’s customer mix and cash receipts before it is treated as durable.

HANMI’s ecological niche is the focused technology supplier that monetizes a narrow, yield-critical step. It takes profit from memory producers’ packaging-capex budgets and from less capable assembly-tool vendors. ASMPT, Hanwha, Besi and captive equipment makers can take that profit pool in different ways: comparable TC tools reduce HANMI’s share and pricing; hybrid tools reduce the useful life of the entire TC category; internal tools reduce accessible third-party spending. The company becomes stronger if future HBM requires more precise but still thermal-compression-based bonding. It becomes weaker if customers standardize hybrid bonding before HANMI earns a production qualification.

Governance is stable but concentrated. Kwak Dong-shin and related parties controlled 55.70% of shares at the end of 2025. Concentrated ownership supports long-term product investment and protects against hostile short-term pressure, but it also limits the influence of minority shareholders. The board structure disclosed around the 2026 meeting included only one outside director, despite full attendance. Related-party transactions were modest in absolute terms, such as KRW 682.7 million of related-party property purchases and KRW 116 million of fees in 2025, but family control and a thin independent board justify a governance discount.

The latest audit opinion was unqualified. Revenue cutoff was a key audit matter because recognition depends on customer-specific delivery and acceptance. One lawsuit with a claimed amount of KRW 1.2 billion was outstanding at year-end, an immaterial amount relative to equity unless it signals broader intellectual-property exposure. No evidence in the reviewed filings indicates a material accounting investigation or auditor qualification.

Current fundamentals and valuation

The last four quarters show a business whose reported earnings are governed by acceptance timing and product mix.

Q3 and Q4 figures are reported results; Q2 2026 is a preliminary fair disclosure that did not include the complete balance sheet, cash-flow statement or customer mix.

The fourth-quarter and first-quarter trough followed delayed HBM4 equipment timing and a lower mix of high-margin TC bonders. Then the second quarter more than reversed the decline, setting company records for quarterly revenue and operating margin. Calculated from the disclosed quarters, first-half revenue was approximately KRW 302.0 billion and operating profit KRW 138.8 billion. Both remain below the first half of 2025, which on the same derivation produced approximately KRW 327.5 billion of revenue and KRW 156.0 billion of operating profit, so the half year is down 7.8% and 11.0% respectively. A single record quarter does not establish a new run rate. Nor did the first-quarter trough establish structural collapse.

Secondary reporting attributed part of the second-quarter strength to Micron and Chinese demand, but the preliminary disclosure did not provide customer-level evidence. The claim remains plausible rather than verified. The half-year filing will be important because a fall in the top-customer percentage and better geographic diversification would make the record quarter higher quality than one driven primarily by delayed SK hynix acceptance.

The order history offers a cleaner signal than quarterly commentary, provided dates are handled correctly.

The apparent KRW 358.7 billion “2026 cumulative” figure is a recycled June 2024 cumulative announcement. Likewise, KRW 42.8 billion belongs to May 2025. The corrected 2026 figure does not capture all sales to SK hynix: contracts below disclosure thresholds, overseas sales and previously announced orders may contribute to revenue. It is simply the defensible total of the two identified 2026 SK hynix contract disclosures.

The latest secondary consensus aggregation placed 2026 revenue near KRW 955.3 billion and EPS near KRW 4,144. This is above the early-2026 market figure of roughly KRW 801 billion revenue and KRW 401.5 billion operating profit cited in broker coverage, but estimates have moved quickly around the first-quarter miss and second-quarter beat. To reach KRW 955.3 billion, HANMI would need approximately KRW 653.3 billion of second-half revenue, or about KRW 326.7 billion per quarter. That is 30% above the record second quarter, making further tool acceptance and overseas orders essential.

The market currently trades the combination of AI-memory capital spending, extraordinary margins and an expectation that TC bonding lasts longer than hybrid-bonding advocates assume. Real fundamental evidence comes from the record Q2, the net-cash balance sheet, existing customer production experience and continued SK hynix orders. The narrative component is the assumption that those economics scale smoothly through HBM5/HBM6 and that HANMI wins enough hybrid business to preserve its terminal value.

Historical valuation offers limited comfort because the business mix changed so sharply. Before the HBM ramp, HANMI commonly traded as a cyclical equipment company. At the July 31 close, audited 2025 multiples were approximately:

§ Based on the cited secondary consensus, not company guidance.

The price/sales ratio looks extreme for machinery because HANMI’s operating margin is extreme. Even after recognizing that distinction, a 51.8 times forecast P/E requires several more years of growth. The peer comparison provides context rather than validation: ASMPT traded near 50 times trailing earnings and Besi around 71 times, so the entire advanced-packaging group carried a substantial AI premium. Hanwha Aerospace’s 29 times multiple is dominated by defense. Expensive peers do not make HANMI cheap.

Cash-flow passthrough changes the valuation basis. The five-year operating-cash-flow/net-income ratio was only 0.69, mainly because the 2023 net-income figure included a very large investment gain. In 2025, operating cash flow exceeded net income, but reported net income still included securities gains. Maintenance capex appears close to current depreciation and amortization of roughly KRW 12 billion, while the remainder of the KRW 74.6 billion property capex and KRW 19.1 billion intangible investment was predominantly growth-related. Normalized owner earnings should start from after-tax operating profit, add non-cash depreciation and deduct maintenance capex, rather than use headline net income or total growth capex without adjustment.

On this basis, audited 2025 owner earnings were roughly KRW 185–200 billion after normalizing investment gains and maintenance spending. The corresponding current multiple is approximately 102–111 times, somewhat higher than the headline P/E. The gap is only 7% to 16%, so the headline P/E remains usable, but owner earnings are the cleaner basis for the scenario analysis because growth capex and securities gains are unusually large.

The absolute valuation below uses 2028 normalized owner earnings and a three-year realization period. Revenue assumptions are materially below management’s former KRW 2 trillion target in the conservative and base cases. The multiples remain generous relative to conventional machinery because HANMI has net cash, high margins and a scarce advanced-packaging position; using ordinary industrial-equipment multiples would produce much lower values.

This is valuation-scenario analysis within a research framework, not investment advice.

The conservative case does not assume collapse. It assumes 2028 revenue above the 2025 record, but a lower operating margin as competition reduces price and product mix normalizes. The base case requires revenue to double from 2025 and owner earnings to increase roughly 80% from normalized 2025 levels. The optimistic case requires successful expansion to multiple memory customers, sustained near-peak margins and credible participation in the next bonding architecture. At the current price, the market sits above base value and below optimistic value.

The expectation gap is concentrated in four metrics. H2 2026 revenue must exceed the Q2 record on average to meet the latest revenue consensus. Operating margin must hold near the mid-40s even as ASMPT and Hanwha gain volume. Customer concentration must fall because of genuine Micron or other overseas growth, and HANMI must convert hybrid-development spending into a customer qualification. A quarterly revenue beat created by delayed acceptance is less valuable than a production order from a second major customer.

The next earnings print will be judged less by reported revenue than by the source of revenue, order timing and margin sustainability. A Q3 result around KRW 239 billion, the secondary consensus indication, would represent a strong year-on-year quarter but a sequential decline from Q2. A larger result with lower margin could indicate price or mix pressure. A smaller result accompanied by large new orders might be more constructive than the income statement alone suggests.

The margin-of-safety test produces a clear answer.

At KRW 214,500, the shares trade at more than twice the conservative value of KRW 102,000. The margin of safety to that case is zero.

The most fragile base-case assumption is KRW 360 billion of sustainable owner earnings combined with a 50 times multiple. Reducing owner earnings to 70% of the assumption gives KRW 252 billion, or about KRW 2,644 per share. At the same multiple plus net cash, value falls to roughly KRW 135,000. Reducing the multiple as well would lower it further.

If earnings remain flat for three years, the current audited earnings yield is about 1.05%, and the dividend yield is well below 1%. Their combined cash-equivalent return is below the Korean 10-year government-bond yield of 4.262% on July 31, 2026. There is no margin of safety at this buy price.

This is a good operating franchise at a price that requires continued exceptional results. Existing holders can rationally value the option on Wide and hybrid bonding, but a new buyer receives little protection against ordinary competitive normalization.

Margin-of-safety sufficiency verdict: none.

Risks, catalysts, and tracking

The highest-probability permanent-loss path is customer and supplier diversification. Probability is medium-to-high and impact high. Watch the top-customer percentage, SK hynix orders awarded to ASMPT and Hanwha, and HANMI’s average selling price. The path is direct: fewer tools reduce revenue, under-absorption lowers operating margin, and evidence that HANMI no longer controls a bottleneck compresses the valuation multiple. The 2025 top-customer share of 43.96% and reports of competitor placements show that this risk is already active rather than theoretical.

Technology substitution has medium probability and high impact. A rapid move to hybrid bonding beginning with HBM4E or HBM5 would reduce the useful life of TC-bonder designs. Follow customer qualification announcements, hybrid-bonder orders, yield disclosures and whether memory makers use hybrid only for development or for volume production. HANMI’s loss path would combine lower TC orders with high hybrid-factory spending and no replacement revenue. Besi and Hanwha already have tools in customer-development channels, while HANMI’s production qualification has not been publicly confirmed.

A slower hybrid transition can also create risk if the market has overestimated total equipment demand. Customers may use improved TC bonding for longer while requiring fewer machines than expected because throughput rises or stack architecture changes. Wide dies, lower stack counts and better utilization could reduce unit demand even if HBM bit shipments increase. Investors should distinguish HBM revenue growth from bonder-unit growth.

Order and acceptance volatility has high probability and medium-to-high impact. Q1 and Q2 2026 provide the evidence: revenue moved from KRW 50.9 billion to KRW 251.1 billion and operating margin from 16.6% to 51.9%. A delayed customer acceptance does not necessarily destroy long-term value, but repeated delays can signal qualification problems, weaker end demand or excess installed capacity. The share price’s high multiple turns temporary timing misses into large de-ratings.

Valuation compression has high probability and high impact. The current market value equals about 95.5 times audited earnings, while the Korean 10-year government bond yielded 4.262% on July 31. A shift from AI scarcity to ordinary capital-equipment pricing can reduce the multiple even while earnings grow. Besi’s valuation has similarly risen far above its ten-year average, showing that this is a sector-wide advanced-packaging premium rather than a HANMI-only phenomenon.

Inventory and growth-capex risk have medium probability and medium impact. Inventory was KRW 150.4 billion at year-end and the inventory-valuation adjustment reached KRW 28.9 billion. Property and intangible capex totaled almost KRW 94 billion in 2025. If product specifications change before delivery, inventory write-downs and unused capacity would reduce cash flow before the income statement fully reflects the deterioration.

Governance risk has low-to-medium probability and medium impact. Family-related parties control a majority, board independence is limited and related-party dealings exist, although reviewed amounts were modest and the audit opinion was clean. The likely damage runs through capital allocation, related-party pricing or aggressive buybacks at elevated valuations, not through immediate insolvency. Minority shareholders depend on the controlling family’s judgment during the technology transition.

Geopolitics is a secondary but real risk. HANMI sells predominantly into Asia, while customers operate global supply chains subject to U.S., Korean, Chinese and allied export controls. Restrictions on advanced semiconductor equipment can reduce access to Chinese customers or require product modifications. The HBM franchise is partly insulated because its largest identifiable customers are allied-country memory producers, but legacy placement and assembly tools may have broader China exposure.

Positive catalysts include additional large Micron or SK hynix orders, disclosure showing a lower top-customer percentage, Wide TC Bonder production qualification, a named hybrid-bonder customer, continued operating margin above 45%, and successful conversion of the Q2 surge into operating cash flow. A Samsung qualification would be particularly consequential because it would challenge the view that HANMI’s success depends on one customer ecosystem.

Negative catalysts include a Q3 revenue relapse without compensating orders, operating margin below 35%, customer acceptance delays, competitor HBM4E or HBM5 wins, slippage of the Wide launch beyond early 2027, absence of a hybrid qualification despite the new facility, a material patent loss, or inventory growing faster than orders.

The expected November 16 earnings date comes from a market calendar rather than a company-confirmed IR notice. Korean statutory filing schedules can shift, so DART should remain the authoritative source as the date approaches.

Revenue and margin should be read together. High revenue with a margin below 35% would indicate weaker mix, price concessions or rising support cost. Low revenue with expanding disclosed orders may merely reflect timing. Low revenue, low orders and rising inventory would be a far more serious combination.

Customer concentration should be tracked from audited or quarterly notes, not inferred from press coverage. Named SK hynix contracts help assess one account, while overseas orders often lack equivalent visibility. The first evidence of durable diversification will be a lower anonymous-customer percentage accompanied by rising export receipts and stable margins.

Hybrid-bonder progress should be measured by customer milestones: evaluation delivery, qualification, production order and revenue. Product unveiling, factory completion and capacity claims are preliminary inputs. A production order matters because it proves that HANMI’s process knowledge transferred beyond thermal compression.

Cross-synthesis and final conclusion

Looking vertically, HANMI has proven one capability over four decades: it can move up the packaging-equipment complexity curve by combining mechanical precision, vision, thermal control and customer-specific production engineering. The company began by replacing imported molds, built a globally competitive placement platform, then entered a process whose economics became central to AI memory. Its success was not pure luck. The founder’s manufacturing background, internal engineering, customer support and willingness to invest before a market became obvious all mattered.

Era tailwinds amplified that capability. Revenue did not climb steadily from 1980 to the present. It contracted sharply in conventional semiconductor downcycles, and the 2024 earnings step-change coincided with an unusually concentrated HBM investment wave. The financial record separates capability from cycle: HANMI maintained attractive margins before HBM, but revenue rose 251.5% in 2024 because customer spending moved suddenly into a tool category where it had an established position. Management created the option; AI demand and SK hynix’s early HBM leadership monetized it.

Those success factors remain present but are changing shape. Precision engineering and an installed base still matter. The balance sheet can finance development without debt. Customer experience reduces qualification risk. The missing element is exclusivity. SK hynix’s vendor diversification and the arrival of ASMPT and Hanwha show that high early share invited competition. HANMI’s historical HBM3E position cannot be carried forward by assertion.

Horizontal comparison clarifies the strategic trade-off. HANMI is more focused and has recently produced higher operating margins than ASMPT. ASMPT has a broader customer base and can sell both TC and hybrid platforms. Besi shows more visible direct-hybrid adoption but benefits when the technology arrives quickly; HANMI benefits when TC remains viable. Hanwha has weaker standalone disclosure but strong industrial funding and Korean customer access. Samsung’s SEMES adds a captive route that does not need to maximize third-party shareholder returns. HANMI occupies the most profitable narrow niche, while competitors surround it from breadth, successor technology, domestic industrial scale and customer integration.

The real advantage is speed and production know-how within TC bonding. The real weakness is that the advantage attaches to a process generation. HANMI can overcome that weakness in two ways. It can extend TC bonding through Wide, fluxless and more precise Griffin platforms, preserving the installed-base economics. It can also win a production qualification in hybrid bonding, converting customer relationships into a successor franchise. Either route can sustain value. The optimistic case requires both: a longer TC runway and enough hybrid credibility to prevent terminal-value collapse.

Current valuation is pre-spending future success. The market capitalization of KRW 20.44 trillion is about 35 times 2025 revenue and 95.5 times 2025 net income. The base valuation already assumes that revenue approximately doubles by 2028, operating margin remains above 40% and the market continues to award a 50 times owner-earnings multiple. Those are not ordinary assumptions for a capital-equipment manufacturer. They reflect a belief that HANMI is a scarce technology franchise.

The market may be underestimating the duration of TC bonding. Hybrid bonding’s surface, alignment, cleanliness and yield requirements are demanding, and repeated reports of delayed mass adoption support HANMI’s Wide strategy. Besi’s shares have at times reacted negatively to signs that memory customers could postpone hybrid adoption. Every delayed generation creates more opportunity for HANMI to sell upgraded TC tools and fund its successor platform.

The market may simultaneously be overestimating HANMI’s share of that duration. ASMPT’s bookings, SK hynix supplier diversification and Hanwha’s qualification activity indicate that TC demand will not accrue to one vendor. Industry TAM growth and company revenue growth are different variables. A 13% market CAGR accompanied by a fall from a historically dominant share to a balanced multi-vendor structure could produce much less than the revenue embedded in the current price.

Over the next year, the decisive variables are H2 revenue acceptance, operating margin, order disclosure and customer mix. HANMI needs quarterly revenue materially above the Q2 2026 level on average to reach the latest secondary consensus. And it must do so without price concessions that pull margin below the low-40s. The most valuable disclosure would be evidence that Micron or another overseas customer contributes enough volume to reduce dependence on the largest anonymous account.

Over three years, three things decide it: Wide TC Bonder qualification, hybrid-bonder production orders and capital discipline. A Wide tool that ships in volume while hybrid remains limited to development could extend extraordinary TC economics. A hybrid qualification at a Tier-one memory maker would preserve terminal optionality. Spending KRW 100 billion on a plant without customer acceptance would reveal that manufacturing capacity was built ahead of process credibility.

Over five years, HANMI’s fate depends on whether it remains a bonding-platform company or reverts to being one of several assembly-equipment vendors. The long-term market is unlikely to reward a single-generation TC specialist at a 50-times multiple. It could reward a company that owns customer relationships across TC, fluxless and hybrid processes, especially if inspection and large-die tools expand content per HBM line.

The company becomes a better investment under one of two conditions. The first is price: a large discount that protects the buyer against margin normalization and supplier diversification. The second is evidence: customer diversification and hybrid qualification sufficiently strong to raise conservative owner-earnings estimates. At the current price, evidence must improve because the valuation provides little protection.

The research judgment should be overturned positively if HANMI earns a named or unmistakably identifiable hybrid-bonder production order, reduces the largest-customer share below 35%, sustains operating margin above 45% while ASMPT and Hanwha add capacity, and converts earnings to cash without further inventory buildup. It should be overturned negatively if operating margin falls below 35% for two quarters, the Wide program slips beyond 2027, no hybrid qualification appears by the end of 2027, or the top customer remains above 50% while competitors receive the majority of new HBM tools.

Bull reasons:

• The audited 2025 operating margin was 43.6%, and Q2 2026 reached 51.9%, confirming exceptional pricing and operating leverage when high-margin bonders are accepted.
• HANMI ended 2025 with KRW 276.2 billion of cash and negligible financial debt, giving it the resources to develop Wide and hybrid tools through a qualification cycle.
• TechInsights’ dated, value-based estimate still showed 71.2% of the defined HBM TC-bonder market through Q3 2025, indicating that the installed production position is substantial even after correcting the headline’s scope.
• Delayed hybrid adoption can extend TC bonding into additional HBM generations, directly supporting the Wide TC Bonder strategy.
• The top-two customer contribution fell from 73.47% in 2024 to 58.78% in 2025, an initial sign that revenue concentration can improve.

Bear reasons:

• The largest anonymous customer supplied 43.96% of 2025 revenue, leaving profit exposed to one customer’s capex, qualification and acceptance schedule.
• SK hynix has qualified Hanwha and reportedly purchased meaningful HBM4 TC-bonder volume from ASMPT, showing that HANMI’s early share is not locked in.
• HANMI has no publicly confirmed volume hybrid-bonder position of record, while Besi, Applied Materials and Hanwha already have tools in customer-development programs.
• The current price equals about 95.5 times audited earnings and 132.7 times audited free cash flow, requiring sustained exceptional growth and margins.
• Management’s 2024 targets of KRW 1.2 trillion revenue in 2025 and KRW 2 trillion in 2026 were far above the realized 2025 result and current trajectory, reducing confidence in long-range numerical claims.

The first pre-mortem script is supplier diversification plus margin normalization. During 2027, SK hynix assigns most incremental HBM4E tools to ASMPT and Hanwha, while Micron dual-sources. HANMI’s 2028 revenue stops near KRW 750–800 billion instead of passing KRW 1 trillion. Competitive pricing and lower factory absorption reduce operating margin from above 40% to around 30–34%. Normalized owner earnings settle near KRW 180–210 billion, and the market compresses the multiple from more than 50 times forward earnings to 30–35 times. Including net cash, the share price could fall toward KRW 60,000–80,000, roughly 63–72% below the July 31 close.

The second script is a failed technology handoff. Hybrid bonding enters volume HBM production in 2028, Besi and Hanwha secure the first major memory production orders, and HANMI’s KRW 100 billion hybrid facility remains underutilized. Wide TC Bonder demand peaks after only one generation. Inventory write-downs rise, capital expenditure remains elevated and free cash flow falls below accounting profit. Even if legacy Vision Placement and TC tools remain profitable, the market reclassifies HANMI from an AI bottleneck to a cyclical assembly-equipment vendor at 20–30 times earnings. A simultaneous earnings and multiple decline can also halve the stock.

HANMI is a high-quality equipment operation embedded in a highly concentrated and technologically transitional revenue base. Its historic engineering record, net cash, customer qualification and margins deserve a premium. The present share price asks the investor to pay for continued TC-bonder relevance, successful customer diversification and at least partial hybrid-bonding success before those outcomes are fully evidenced.

At KRW 214,500, the shares sit near the top of the base hold band and above the base-case point value. Existing holders retain meaningful upside if the optimistic scenario materializes, but expected base returns are slightly negative over three years and the conservative downside is severe. The most important concern is the combination of a single-generation moat and a multi-generation valuation.

【Company-profile scores】

• Fundamental quality: high
• Growth: high
• Moat: medium
• Financial soundness: strong
• Management credibility: medium
• Valuation attractiveness: low
• Risk level: high
• Suitable investor type: long-term growth investors able to tolerate semiconductor-capital-equipment cycles

【Investment rating】

• Rating: Hold
• One-line thesis: Exceptional TC-bonder economics are capitalized; customer diversification and hybrid-bonder qualification are needed before risk-adjusted upside improves.
• Ideal buy price:

【Ideal Buy Price】75,000–82,000 KRW

This is 19.6% to 26.5% below the KRW 102,000 value implied by the conservative scenario and would provide protection against supplier diversification and margin normalization.

• Acceptable hold price: KRW 165,000–220,000, approximately ±15% around the KRW 192,000 base value
• Clearly overvalued price: KRW 350,000–380,000, beginning above 110% of the KRW 314,500 optimistic value
• Current-price classification: acceptable hold
• Whether to wait for a better price: yes; a new purchase requires the KRW 75,000–82,000 ideal-buy range, unless verified customer diversification and a production hybrid qualification materially raise the conservative value. The opportunity cost is missing a rapid rerating if both arrive before a correction.
• Target holding horizon: 3–5 years
• Expected annualized return: conservative −21.9%, base −3.6%, optimistic 13.7%, before dividends and taxes
• Max-loss risk: approximately 60–70% if HBM customers shift incremental tools to competitors, operating margin normalizes near 30%, and the valuation falls toward ordinary equipment multiples
• Reassessment-trigger signals: operating margin below 35% for two consecutive quarters; largest-customer share above 50%; no Wide qualification by end-2027; no hybrid production qualification by end-2027; inventory above 35% of trailing revenue; or a top-two customer share below 45% accompanied by stable margins, which would support a positive reassessment

【Valuation Range】

• current: 214,500 KRW (close as of 2026-07-31)
• bear (conservative · ideal buy zone): [75,000, 82,000]
• base (fair · acceptable hold zone): [165,000, 220,000]
• bull (optimistic · above the clearly-overvalued line): [350,000, 380,000]

The principal research uncertainties are the identities behind the filing’s anonymous customer percentages; the proprietary scope and methodology of the TechInsights market-share estimate; the absence of full cash-flow and customer data in the preliminary Q2 release; confidential customer qualification schedules for Wide and hybrid tools; and the limited, fast-changing analyst-consensus sample. These blind spots make order disclosures, DART filings and customer procurement evidence more reliable than headline market-share or target-price figures.

The primary source hierarchy used here was HANMI’s DART/KRX annual and quarterly filings, audited notes, shareholder-meeting materials and company releases; official ASMPT and Besi results; customer and competitor disclosures; and dated market-price data. Secondary industry reporting was used where customer qualifications or equipment allocation were not disclosed directly, and those points have been described as reported rather than established facts.

Other tickers mentioned

• 0522.HK — ASMPT is the closest listed head-to-head competitor across thermal-compression and hybrid bonding.
• BESI.AS — Besi is the principal listed reference for direct hybrid-bonding equipment and adoption timing.
• 012450.KO — Hanwha Aerospace owns Hanwha Semitech, a Korean TC- and hybrid-bonder challenger.
• 000660.KO — SK hynix is HANMI’s most visible HBM customer and the main disclosed order-flow signal.
• 005930.KO — Samsung Electronics is a major HBM producer whose captive and external equipment choices affect HANMI’s accessible market.
• MU.US — Micron is the most important visible route toward reducing HANMI’s dependence on SK hynix.
• 2330.TW — TSMC’s advanced-packaging activity influences hybrid-bonding adoption and equipment demand.
• AMAT.US — Applied Materials participates in integrated hybrid-bonding development systems.
• ASM.AS — ASM International is a significant shareholder in ASMPT and a relevant front-end equipment reference.
• NVDA.US — Nvidia accelerator roadmaps drive HBM qualification schedules and memory-capacity demand.

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