Introduction
The KGD testing market is moderately consolidated at the equipment level and more fragmented at the level of interface hardware and services. Automatic test equipment for logic, AI and memory devices is supplied mainly by two companies, Advantest and Teradyne, whose platforms are installed at nearly every major OSAT, foundry and IDM. Wafer probers come from a small group of Japanese and Taiwanese suppliers, led by Tokyo Electron and Tokyo Seimitsu. Probe cards, which are the most device-specific part of the test cell, are supplied by a wider group that includes FormFactor, Technoprobe, Micronics Japan, Japan Electronic Materials and MPI Corporation, alongside regional suppliers in China and South Korea. Burn-in, handling, contacting and inspection are served by companies such as Aehr Test Systems, Cohu and Chroma ATE.
Competition is based on technical capability, qualification history and the ability to support customers during the ramp of new devices, rather than on price alone. A new AI processor or HBM generation requires a tester configuration, a probe card design and a thermal solution that work together, and chip companies typically qualify only a few suppliers for each element. Once a supplier is qualified, it tends to remain in place for the life of a product family, which gives incumbents an advantage. At the same time, each device generation reopens the competition, because new pitches, power levels and stack heights require new solutions.
A defining feature of the market since 2023 has been the formation of alliances between suppliers of different parts of the test cell. Equity investments, joint development agreements and integrated test cell offerings have linked testers with probe cards and probers, reflecting customer demand for complete, pre-validated KGD solutions for high-power dies.
Key Player Positioning
Leaders are the companies that supply the core of most KGD test cells and hold strong positions across several device types and regions. Advantest and Teradyne lead in test systems for logic, AI processors and memory, with broad installed bases, global service networks and dedicated products for HBM and stacked devices. FormFactor and Technoprobe lead in advanced probe cards: FormFactor has a strong position in DRAM and HBM probe cards and in engineering probe systems, while Technoprobe has grown through high-performance logic and AI probe cards and expanded its scope by acquiring Teradyne's device interface business in 2024. Tokyo Electron is a leader in wafer probers, with its Prexa platform used in high-volume probe testing and now in integrated KGD cells.
Challengers hold strong positions in specific steps or device classes and are expanding their scope. Micronics Japan is a major supplier of memory probe cards and has strengthened its position through its 2025 partnership with Advantest. Tokyo Seimitsu is the second major Japanese prober supplier and is co-developing a die-level prober with Advantest. Cohu supplies test handlers, contactors and inspection systems, and has extended into silicon carbide burn-in and HBM inspection. Chroma ATE supplies SoC, power and system-level test solutions and has a strong customer base in Taiwan and China.
Niche players focus on particular technologies where they hold specialised expertise. Aehr Test Systems is a specialist in wafer-level burn-in for silicon carbide and, increasingly, for AI processors, complemented by package-level burn-in from the Incal Technology acquisition. MPI Corporation supplies probe cards and probers, with a strong position among Taiwanese customers, and Japan Electronic Materials specialises in probe cards for memory devices. These companies are important in their segments but lack the breadth of portfolio and global reach of the leaders.
Competitor Dashboard
The competitor dashboard rates each company on four dimensions: the breadth of its KGD-relevant product portfolio, its regional presence, the depth of its strategic alliances and its industry experience. Ratings are qualitative assessments and are consistent with the leader, challenger and niche grouping. Table 11.1 presents the ratings for the twelve companies covered.
The competitive structure of known good die testing reflects high technical barriers, a concentrated supplier base, and a customer base that is itself concentrated in a few large manufacturers and service providers. The ratings that follow are assessed for the equipment, probe card and burn-in suppliers that make up the market.
Taken together, the forces describe a market in which profitability is supported by technical barriers and recurring consumable revenue but constrained by powerful, concentrated customers. The balance between these influences differs by segment: wafer-level probe testing and known-good-stack testing of HBM are most exposed to buyer concentration, while burn-in and stress screening for AI processors and silicon carbide dies remains a developing field where suppliers have more room to shape specifications. Singulated die-level testing sits between the two, because it requires new handling and thermal capability that only a few suppliers can deliver, yet it serves customers who can compare competing integrated test cells directly.
Threat of New Entrants: Low
The threat of new entrants is Low. A new tester or prober supplier must match decades of accumulated know-how in pin electronics, timing accuracy, thermal control and wafer handling, and must then pass long qualification cycles at foundries, IDMs and OSATs that will not risk yield on unproven equipment. Customers build test programs, correlation data and operator training around a specific platform, and switching to a new platform means re-qualifying test content for every device in production. This installed-base effect protects established tester families and makes it very difficult for an entrant to gain a first production slot.
Probe cards look more accessible because each card is a custom product and regional suppliers exist in Taiwan, South Korea and China. However, the cards needed for HBM, AI processors and stacked dies require MEMS probe fabrication at very fine pitch, high current per probe and very tight planarity across large areas, and a single card can carry tens of thousands of probes. Capital requirements for MEMS lines, the need for access to scarce multilayer ceramic substrates, and the fact that leading tester makers have taken equity positions in established probe card suppliers since 2023 all raise the barrier further. The main route for new entrants is domestic substitution in China, where policy support and restricted access to some foreign tools encourage local tester and probe card suppliers. This creates new competitors in mature-node and domestic segments, but it does not yet challenge the incumbents in the most demanding AI and HBM applications.
Burn-in and stress screening is the segment where entry is most visible, because wafer-level and package-level burn-in for AI processors and silicon carbide devices is still a developing application and customers are open to evaluating new approaches. Even here, however, the leading positions have been taken by suppliers with existing installed bases or through acquisition: Aehr Test Systems added package-level burn-in for AI processors by buying Incal Technology in 2024, and Cohu entered silicon carbide burn-in in October 2024 using its established handler and contactor business. A start-up would need to offer a clear improvement in cost per screened die or in thermal capability, and would still face the reluctance of customers to place unproven equipment in a high-value production flow. Software-led entrants, such as analytics firms that optimise test content using machine learning, face lower capital barriers, but their value depends on integration with the data streams of established testers, which again favours partnerships with incumbents over independent entry.
Bargaining Power of Suppliers: Medium
The bargaining power of suppliers is Medium. On the one hand, several inputs are highly concentrated. Multilayer ceramic space transformers, specialised probe alloys, high-current pin-electronics components and advanced thermal control units are supplied by a short list of firms, and probe card makers have reported substrate availability as a constraint during periods of rapid HBM demand growth. Fluorinated heat-transfer fluids used in some thermal sub-systems face supply changes as 3M completes its exit from PFAS manufacturing by the end of 2025, which forces equipment makers to qualify alternatives.
On the other hand, the leading equipment makers design many critical components in-house, including tester ASICs and probe technologies, which limits the leverage of outside suppliers. Long-term supply agreements and dual sourcing are common, and the largest participants have the scale to secure priority allocations. Supplier power is therefore felt mainly by smaller probe card and handler makers that lack in-house substrate or component capability, and by all participants during demand spikes, rather than as a constant structural pressure.
The equity links formed between tester makers and probe card suppliers also change the supplier relationship. Teradyne's 10% holding in Technoprobe, completed in May 2024, and Advantest's stakes in FormFactor, Technoprobe and Micronics Japan, announced in January and February 2025, give the largest system suppliers closer visibility of probe card roadmaps and capacity. For the tester makers, probe card suppliers are partners as much as vendors, and the alliances reduce the risk that a shortage of qualified cards delays tester shipments. From the perspective of an OSAT or foundry, the probe card supplier is itself a supplier with meaningful power, because a card designed for a specific HBM or AI processor die cannot easily be sourced elsewhere at short notice. When a new device ramps, the first qualified card supplier often holds a strong position until a second source completes qualification, which can take several months. This device-specific dependency is the main reason supplier power is assessed as Medium rather than Low.
Bargaining Power of Buyers: High
The bargaining power of buyers is High. Demand for the most advanced KGD test equipment comes from a small number of very large organisations: leading foundries with advanced packaging lines, the three main HBM producers, a handful of large OSATs and test houses, and the power device IDMs. Each of these buyers places large, programme-level orders and can concentrate a significant portion of a supplier's annual revenue in one decision. Buyers run competitive evaluations between tester platforms and between probe card vendors, request multi-sourcing to protect supply continuity, and negotiate on cost per tested die rather than on equipment price alone.
Buyer power is reinforced by the fact that chip owners, including fabless companies and hyperscalers designing their own accelerators, increasingly specify the test strategy and influence which platforms their OSAT partners use. Advantest's stated aim, when taking stakes in FormFactor and Technoprobe in January 2025, of offering customers several viable probe card sources reflects this pressure for choice. Switching costs moderate buyer power once a platform is in production, because re-qualifying test programs is slow and expensive. However, buyers use the start of each new device generation, such as the move from HBM3E to HBM4 or from one accelerator family to the next, to re-open supplier selection, which keeps pricing and service terms under steady negotiation.
The concentration of demand is particularly marked in the fastest-developing parts of the market. Known-good-stack testing of HBM is purchased by only three memory makers, and the leading AI accelerators are assembled by a small number of foundry and OSAT partners, so a single programme decision can determine the winner of a large share of the most advanced probe card and tester demand for a full product cycle. In power devices, the shift to silicon carbide and gallium nitride modules is led by a limited group of IDMs that operate their own test floors and develop much of their test know-how internally. Test service providers such as KYEC and ASE add a further layer of buyer power, because they aggregate demand from many chip owners and can standardise on preferred platforms across their sites. Smaller buyers, including fabless companies with niche RF, MEMS or industrial devices, have far less influence, but they account for a smaller portion of spending on advanced KGD equipment.
Threat of Substitutes: Low
The threat of substitutes is Low. Known good die testing has no direct functional substitute when multiple dies are assembled into a single package, because a defective die discovered after assembly destroys the value of every other die and of the interposer or substrate. Alternatives exist mainly in where and how much testing is done. A chip owner can rely more on package-level final test and system-level test, accepting higher assembly scrap in exchange for less wafer and die test, or can depend more on built-in self-test and redundancy, such as spare memory rows and lanes that can be repaired after assembly.
These approaches reduce the intensity of some KGD steps for lower-value devices, consumer products and designs with generous repair capability, but they do not remove the need for pre-assembly screening of high-value logic, HBM and power dies. Advances in design-for-test and in-field monitoring change the mix of test methods rather than replacing KGD testing. Improvements in wafer fabrication yield can also reduce the number of failing dies, yet even a low defect rate multiplied across many dies in a stack or module makes screening economically necessary.
There is also a partial substitution between test steps within the KGD flow rather than between KGD testing and an outside alternative. Wafer-level burn-in can replace part of package-level burn-in, die-level testing of singulated dies can reduce the need for some wafer-level test, and system-level test of a completed module can catch defects that escape structural tests. These shifts move spending between segments of the market but generally keep it within the market, because the equipment and probe card suppliers serve several steps. The most credible long-term substitute pressure comes from design techniques that make dies self-testing and self-repairing, such as more extensive built-in self-test, redundant lanes on die-to-die interfaces and in-field health monitoring. These reduce the duration of external test for some devices, but they still require probe access at wafer level to activate and verify the on-die structures, and they do not remove the need for burn-in to screen early-life failures in automotive and data-center parts.
Intensity of Competitive Rivalry: High
Competitive rivalry is High. Advantest and Teradyne compete directly for tester sockets in logic, memory and power test; Tokyo Electron and Tokyo Seimitsu compete in wafer probers; and FormFactor, Technoprobe, Micronics Japan, Japan Electronic Materials and MPI Corporation compete for probe card business that is won or lost device by device. Rivalry is driven by technology rather than by price alone: suppliers compete on probe pitch and current capacity, thermal control at high power, parallelism in memory test and the speed of test-cell bring-up.
The pace of product introductions shows the intensity of competition. Teradyne launched the Magnum 7H for HBM in August 2025, Advantest and Tokyo Seimitsu announced a die-level prober development in December 2025, and Teradyne and Tokyo Electron introduced an integrated KGD test cell in June 2026. Burn-in suppliers, including Aehr Test Systems and Cohu, compete to move stress screening from package level to wafer level for AI processors and silicon carbide devices. Equity alliances between testers and probe card makers add a strategic dimension, as each supplier group tries to secure qualified positions in the most important AI and HBM programmes. Cyclicality adds pressure: in periods when memory or consumer device demand weakens, suppliers compete hard for a smaller pool of orders.
Rivalry also takes place across geographic and customer boundaries. Japanese suppliers have long held strong positions with domestic memory makers and probers, U.S. suppliers with logic and AI processor programmes, and Taiwanese and Korean suppliers with local test houses and memory producers. These positions are being contested as AI accelerator and HBM programmes cross regions, and as Chinese customers develop domestic supply chains for testers and probe cards. Service quality is a competitive factor in its own right: suppliers maintain applications engineers on customer sites, offer test program conversion between platforms, and compete on the speed of probe card repair and re-tipping. Because the number of large programmes is limited, losing a qualification at a leading foundry or memory maker can remove a supplier from a device generation for several years, which raises the stakes of each competitive evaluation and encourages aggressive commercial terms during new product introductions.
Threat of New Entrants: Low; Platform know-how, long qualification cycles, installed test programs and scarce MEMS probe and substrate capability deter new suppliers.
Bargaining Power of Suppliers: Medium; Ceramic substrates, probe alloys and thermal sub-systems are concentrated, but leading equipment makers design many critical components in-house.
Bargaining Power of Buyers: High; A few foundries, HBM producers, OSATs and power IDMs place large programme orders and demand multi-sourcing at each new device generation.
Threat of Substitutes: Low; No functional alternative exists to screening high-value dies before multi-die assembly; only the mix of test steps can shift.
Intensity of Competitive Rivalry: High; Testers, probers and probe card makers compete device by device on pitch, thermal control and parallelism, with frequent product launches.
