Introduction
The thermocompression bonding equipment market has a two-tier structure. Production-grade, fully automatic TC bonders for HBM stacking and for large AI logic packages are supplied by a small group of companies, and a few of them account for most installed tools at the three DRAM makers and at the leading foundry and its outsourced assembly partners. Beneath that tier, a wider and more fragmented group of suppliers serves flip-chip, sensor, photonics, university and pilot-line customers with semi-automatic and manual bonders. The market is therefore consolidated by value and fragmented by number of suppliers.
Competition in the production tier is shaped by qualification rather than by list price. A new bonder must pass a customer's process qualification for a specific product, such as a 12-high or 16-high HBM stack or a given interposer size, before it can be ordered in volume, and that qualification can take several quarters. Once qualified, a supplier tends to keep the position for the life of the product generation, which explains why order shares change mainly at generation transitions. The transition from HBM3E to HBM4 during 2025 and 2026 opened such a window. SK hynix, which relied on Hanmi Semiconductor as its main TC bonder supplier from 2017, added ASMPT toward the end of 2024 and Hanwha Semitech in March 2025, and in June 2026 placed a KRW 44.2 billion order with Hanmi for its upgraded TC Bonder 4.5 Griffin. Samsung Electronics, which has used its affiliate SEMES and Japan's Shinkawa for TC-NCF bonders, was reported in January 2026 to be discussing additional TC-NCF and fluxless bonders with ASMPT.
The main dimensions of competition are placement accuracy, the number and speed of bond heads, force and temperature uniformity across larger dies, fluxless oxide-removal capability, field service near the customer's line, and protection of intellectual property. Patent disputes, such as Hanmi's December 2024 suit against Hanwha Semitech and Hanwha's countersuit in 2025, show that incumbents defend their process know-how actively. Suppliers are also positioning for hybrid bonding: Besi leads that segment, Hanwha Semitech announced its second-generation hybrid bonder in February 2026, and Hanmi has said it plans to launch a hybrid bonder by the end of 2027. A credible hybrid bonding roadmap increasingly influences which TC bonder supplier a customer selects, because buyers prefer partners that can support both processes on the same lines.
Key Player Positioning
Leaders. Hanmi Semiconductor and ASMPT form the leading group. Hanmi has the deepest installed base in HBM stacking, supplies both SK hynix and Micron, was named a Top Supplier by Micron in 2025, launched its HBM4-dedicated TC Bonder 4 in May 2025 and reported record sales of KRW 576.7 billion for 2025. It is extending its range into 2.5D big-die TC bonders, which it showed at SEMICON Taiwan in September 2025, and it broke ground on a seventh plant in Incheon in January 2025. ASMPT is the broadest supplier across applications: it holds the process-of-record position for chip-to-substrate TC bonding at a major OSAT partner of the leading foundry, won repeat chip-to-wafer orders from a leading integrated device manufacturer in June 2026, ships fluxless tools based on its plasma Active Oxide Removal technology, and has become a second source for HBM at SK hynix. Its global sales and service network outside Korea gives it an advantage in logic packaging.
Challengers. Hanwha Semitech, SEMES, Kulicke & Soffa, Besi and Shinkawa form the challenger group. Hanwha Semitech entered HBM production in March 2025, opened a technology centre next to SK hynix's Icheon site in May 2025 and is developing fluxless and larger-head TC bonders, but its customer base is still narrow. SEMES has a protected position at Samsung Electronics and a broad equipment portfolio, but it sells mainly to its affiliate. Kulicke & Soffa was among the first suppliers to run fluxless TC bonding in production and has a strong global presence, while its share of HBM stacking remains small. Besi has a strong position in die-to-wafer bonding through its TCB Next platform and its hybrid bonding leadership, supported by Applied Materials' 9% shareholding acquired in April 2025. Shinkawa, part of the Yamaha Motor group, supplies TC-NCF bonders to Samsung and has long experience in flip-chip bonding but a narrower customer reach.
Niche players. Toray Engineering, Shibaura Mechatronics and Finetech serve specific applications. Toray Engineering and Shibaura Mechatronics supply high-accuracy flip-chip and TC bonders for chip-on-wafer, image sensor and specialty packages, mainly in Japan and other parts of Asia. Finetech focuses on manual and semi-automatic bonders used in photonics, sensor assembly, research institutes and prototyping, where flexibility matters more than throughput. These companies rarely compete for high-volume HBM orders, but they are well placed in photonics and co-packaged optics development, where low volumes and frequent process changes favour configurable tools.
Competitor Dashboard
The dashboard rates each company on product portfolio breadth for TC bonding and related die bonding, regional sales and service presence, strategic alliances with customers and partners, and industry experience in semiconductor assembly equipment. The ratings are consistent with the leader, challenger and niche grouping.
The structure of competition in thermocompression bonding equipment is shaped by three features: a technology that is difficult to master at production scale, a small group of very large buyers, and a credible alternative bonding method that is moving from development toward production. The five forces are assessed in that context.
The assessment focuses on production-grade TC bonders used for memory stacking, logic and chiplet assembly and fine-pitch flip-chip, which account for most equipment spending, while noting where the manual and semi-automatic segment for research and photonics behaves differently. Ratings reflect conditions observed in 2025 and the first half of 2026 and the direction expected over the forecast period.
Threat of New Entrants
The threat of new entrants is rated Medium. Entry barriers are high for the production segment that serves HBM stacking and large logic packages. A new supplier must design bond heads that combine fast thermal ramping with uniform force across large dies, achieve placement accuracy in the low single-digit micrometre range at production throughput, and accumulate the process data that customers require before a tool can be accepted as process of record. Qualification cycles take several quarters, and customers rarely change a qualified tool in the middle of a product generation. Patent portfolios covering bond heads, stacking methods and gang bonding add a legal barrier, as the patent litigation between Hanmi Semiconductor and Hanwha Semitech, which started in December 2024 and continued with a countersuit in 2025, demonstrates.
Barriers are lower than these points suggest in two respects, which is why the rating is not High. First, the market has shown that a well-capitalised firm with adjacent competence can enter in a short period. Hanwha Semitech, drawing on its experience in surface-mount and die-bonding equipment, passed SK hynix's qualification and signed its first TC bonder contracts in March 2025, and ASMPT moved from a marginal position in HBM to a supply role at SK hynix within roughly a year. Second, buyers actively want more suppliers to reduce dependence on incumbents and gain pricing power, which lowers the commercial barrier for capable entrants. In China, policy support for domestic semiconductor equipment and restrictions on access to some foreign technology create space for local toolmakers to develop TC bonders for domestic memory and packaging companies. At the low-volume end of the market, manual and semi-automatic bonders for research and photonics have lower technical and capital barriers, and several regional specialists compete there. The likely entrants are therefore established equipment companies from adjacent fields, rather than start-ups.
Capital needs for entry are moderate compared with front-end wafer equipment, but the cost of building application laboratories, hiring process engineers and sustaining a field service organisation near customer fabs is significant, and a new supplier must fund this effort for several years before volume orders arrive.
Bargaining Power of Suppliers
The bargaining power of suppliers is rated Medium. Toolmakers rely on a small number of specialist vendors for high-precision linear motors and stages, encoders, force sensors, optical modules, heaters and motion controllers. Some of these parts have few qualified sources and long lead times, and a change of supplier can require requalification of the complete bond head or alignment system, which gives these vendors a degree of power. During the period of tight supply for industrial electronics in 2021 and 2022, component shortages extended lead times across the semiconductor equipment industry, and similar risks reappear whenever demand for back-end tools rises quickly.
Several factors limit this power. Leading toolmakers design the most critical parts, including bond heads, heater assemblies and alignment algorithms, in-house, so that the components they buy are often configurable standard products rather than unique designs. Most precision component suppliers also serve much larger markets, such as front-end wafer equipment, machine tools and industrial automation, and TC bonder makers are not their dominant customers, which reduces dependence in both directions. Materials suppliers for non-conductive films, fluxes and underfills hold influence over process choices, but their commercial relationship is mainly with the end user rather than with the toolmaker. Overall, suppliers can delay output and push prices for constrained parts, but they do not control the direction of the market.
A specific source of supplier influence comes from the materials side of the process. When a memory maker changes its non-conductive film, flux or underfill, or moves to a new bump metallurgy, toolmakers must adapt heater profiles, force settings and in some cases chamber design to the new material. Materials suppliers therefore indirectly shape tool requirements and the timing of upgrades. Geographic concentration is a second consideration: many precision motion, optical and sensing components come from Japan, Germany and Switzerland, and changes in trade rules, currency movements or regional disruptions can raise costs for toolmakers located elsewhere. Larger toolmakers respond by qualifying second sources and holding buffer stock for long-lead items, which smaller rivals find harder to fund.
Bargaining Power of Buyers
The bargaining power of buyers is rated High. The customer base for production TC bonders is highly concentrated. Three memory makers account for nearly all HBM capacity, and on the logic side, demand is led by the leading foundry and a few OSAT partners and IDMs. Each of these buyers places orders for tens of tools at a time, defines the technical specification jointly with the supplier and can delay or redirect orders as its own capacity plans change. Their purchasing power has been visible in recent years. In 2025, SK hynix moved from a single main TC bonder supplier to a multi-vendor model by qualifying Hanwha Semitech and expanding orders with ASMPT, and in January 2026, Korean press reports described Samsung Electronics discussing TC bonder supply with ASMPT in addition to its in-house SEMES and Japan's Shinkawa.
Buyers also have options for vertical integration. Samsung already sources a large part of its TC-NCF bonders from SEMES, its own equipment subsidiary, which gives it a benchmark for cost and capability when it negotiates with outside suppliers. Large buyers can also shift volume between process generations, for example by extending existing tools through upgrades or by moving part of future capacity to hybrid bonding, which strengthens their position in price discussions. Switching costs moderate buyer power to some degree, because a qualified tool is embedded in a production process and changing it carries yield risk, and toolmakers with a unique process capability can hold price for a period, as Hanmi Semiconductor's reported price increase to SK hynix in 2025 shows. On balance, the size and concentration of buyers mean that they set most commercial terms.
Buyer power is lower in the fragmented part of the market. Universities, research institutes, photonics start-ups and small specialty manufacturers that purchase manual and semi-automatic bonders buy one or two tools at a time and depend on the supplier for process guidance, so they have limited negotiating strength. Their contribution to total equipment revenue is small, however, so the overall rating remains High. Buyers in China represent a further case: restrictions on access to some foreign equipment and technology narrow their choice of suppliers, which reduces their bargaining position with the vendors that remain available to them while increasing their incentive to support domestic toolmakers.
Threat of Substitutes
The threat of substitutes is rated Medium. The main substitute is hybrid bonding, which joins copper pads and dielectric surfaces directly without micro-bumps or solder. Hybrid bonding removes the gap between stacked dies, lowers stack height, improves thermal performance and supports interconnect pitches far finer than those achievable with bumped processes. Besi, the leading supplier of hybrid bonding equipment, has developed its tools in partnership with Applied Materials since 2020, and Hanwha Semitech announced its second-generation hybrid bonder in February 2026. Samsung Electronics presented hybrid bonding results at NVIDIA GTC in 2026, and SK hynix executives have said that hybrid bonding will be necessary for stacks of 20 layers or more.
Several factors keep the threat at a Medium level rather than High during the forecast period. JEDEC's HBM4 standard, published in April 2025, raised the allowable package height to about 775 micrometres, which allows 16-high stacks using micro-bump TC bonding, and press reports in March 2026 indicated that JEDEC was considering a further relaxation for HBM4E. Hybrid bonding requires front-end-grade surface preparation, cleanliness and planarisation, along with new inspection steps, which raise cost and slow yield learning. For chip-to-substrate assembly of large logic dies on organic substrates, hybrid bonding is not yet a practical replacement. Mass reflow with capillary underfill remains a lower-cost alternative for coarser-pitch flip-chip, but it loses competitiveness as pitch shrinks and dies become thinner and larger. The substitution risk therefore grows over time and is concentrated in the highest-stack memory products and the finest-pitch chip-to-wafer applications.
Toolmakers have responded to this threat by treating hybrid bonding as an adjacent business rather than an outside substitute. Besi, ASMPT, Hanwha Semitech and Hanmi Semiconductor all have hybrid bonding programmes, and Hanmi announced in July 2025 an investment programme that includes hybrid bonder development targeted for launch by the end of 2027. This means that substitution would shift spending between product lines of the same suppliers to some extent, although hybrid bonding also brings front-end equipment makers into the back-end process, which changes the competitive set.
Competitive Rivalry
Competitive rivalry is rated High. Although the number of suppliers capable of high-volume production tools is small, they compete intensely for a few large orders, and the loss or gain of a single customer can shift relative positions quickly. Hanmi Semiconductor has held the leading position in HBM TC bonders, but SK hynix's move to multi-sourcing in 2025 brought Hanwha Semitech and ASMPT into direct competition for the same production lines, and the orders announced through 2025 and the first half of 2026 show volume being split among the three. In logic packaging, ASMPT announced orders for 19 and then 15 chip-to-substrate TC bonders from a major OSAT partner of the leading foundry in December 2025, while Kulicke & Soffa and Besi compete with fluxless and die-to-wafer platforms, and Hanmi entered the 2.5D large-die segment with a big-die TC bonder shown in September 2025.
Rivalry is carried out through technology rather than price alone. Suppliers compete on bond head count and throughput, placement accuracy, fluxless capability, support for larger dies and taller stacks, and the speed of on-site support. Patent litigation, aggressive staffing of customer-adjacent service centres and heavy investment in hybrid bonding as a hedge are all signs of a contest in which each supplier tries to secure a role in the next product generation. The cyclical nature of memory capital spending adds pressure, because suppliers with high fixed costs in engineering and new plants must compete hard for orders during slower periods.
Rivalry also has a strong customer-specific dimension. Because each memory maker runs a different stacking process, a supplier's strength at one customer does not transfer automatically to another, and suppliers compete separately for each process of record. Reported financial results reflect this: ASMPT's TCB revenue rose sharply in 2025 as it gained memory and logic orders, while Hanmi Semiconductor reported record annual sales for 2025 but its lowest quarterly TC bonder sales in two years in the final quarter of 2025, as order timing at memory customers slowed during the transition from HBM3E to HBM4.
Taken together, the five forces describe a market in which concentrated buyers and intense rivalry limit pricing freedom, while high technical barriers protect established suppliers from casual entry. The balance of power is likely to depend on which suppliers secure process-of-record positions for HBM4E and for next-generation chiplet packages.
Threat of New Entrants: Medium; High technical and qualification barriers, but buyers' multi-sourcing has let capable adjacent equipment makers enter quickly.
Bargaining Power of Suppliers: Medium; Specialist precision components have few sources, but toolmakers design critical bond heads and software in-house.
Bargaining Power of Buyers: High; A few memory makers, foundries and OSATs place large orders, set specifications and actively dual-source suppliers.
Threat of Substitutes: Medium; Hybrid bonding targets high stacks and fine pitch, but relaxed HBM height limits and higher cost delay substitution.
Competitive Rivalry: High; A small group of suppliers competes intensely for few large orders through technology, service and patent strategy.
