Teradyne Inc · Moat analysis

Report previewTER.USReport date Last updated
← Back to reports
Report contents
01

Industry landscape

Where the company competes: a high-barrier duopoly in chip test

Teradyne builds the machines that decide whether a finished chip ships or gets scrapped. Semiconductor test is the core of that business. It supplies automated test equipment, or ATE: the tester itself, the handler that feeds parts into it, the interface hardware, and the software that runs the test program. Chipmakers buy testers as capex and keep them for years under service and software contracts, so a socket win at a design start generates follow-on revenue long after the tool ships. Upstream sit precision instrumentation, custom high-speed ASICs, mechanical handling and a large proprietary software stack. Downstream are the customers: fabless designers, foundries, integrated device manufacturers, and outsourced assembly and test houses (OSATs, which package and test chips for other companies).

Submarket definition matters here. Teradyne competes mainly in SoC and compute test, the testers used for processors, AI accelerators and networking silicon, and in memory test, including the DRAM and HBM (high-bandwidth memory) parts that sit alongside AI accelerators. It also sells system-level test, where the packaged device rather than the bare die is exercised, and is building toward test for co-packaged optics. Those are adjacent sockets inside the same customer base, not separate markets.

The arena is global. Test demand clusters where leading-edge fabs sit: Taiwan, South Korea, the United States, Japan and China. Competition is not fought region by region. A tester qualified on one customer's line is sold into that customer's programs worldwide, and qualification is usually tied to a specific chip design. Teradyne's 2025 annual report, filed February 19, 2026, describes a company that also runs a robotics business and a product-test business for defense and aerospace. Both are smaller, and neither changes the answer to the competitive question.

Market structure: two vendors take roughly 85% of the spend
SupplierApprox. ATE shareBasis
Teradyne~30%Company, at Bank of America's Global Technology Conference, June 2, 2026
Advantestroughly halfEstimate; Teradyne does not publish rival shares
Cohulow single digitsMostly handlers and contactors
Chroma, HYC and other Chinese supplierssmall, largely domesticMature-node test in China

Advantest is the larger vendor and the one that sets the tempo in memory and SoC test. Teradyne's own figure of about 30% of the ATE market, given at that June 2026 conference, implies the two leaders split something close to 85% of the spend between them. That is the practical test for entry barriers: the number of suppliers who can win a leading-edge socket fits in one hand, and the top of the list is not crowded.

Qualification cost is why the field stays narrow. A tester has to be designed into a chip program well before volume production, and the customer's engineering team spends months writing and validating the test program against it. A tester that misses defects ships bad parts into systems where the cost of failure dwarfs the price of the tool. Test programs are expensive to port between architectures, and an installed base of thousands of systems creates a service and spares annuity. Customers nonetheless want a second source, and that same impulse caps Teradyne. Management said dual-vendor strategies at major customers hold its share near 30% in the near term, with gradual gains toward a 35% to 38% target over three to four years. Concentration is high and barriers are high; the second-source convention is what keeps either vendor from running away with the market.

The size of the prize is changing. Teradyne put the ATE market at $12 billion to $14 billion and said it could reach $20 billion by the end of the decade if wafer fab equipment spending reaches $250 billion to $300 billion. That is a company projection tied to a capex path, not a reported result. The mechanism is test intensity: advanced packaging and HBM stacks require more test insertions per wafer, so test spending grows faster than unit volumes. Management also projected co-packaged optics test at $100 million in 2026, rising to $300 million to $700 million by 2028 and above $1 billion by 2030, with Teradyne holding strong share in network-switch test and expecting a two-vendor structure. That submarket is a rounding error against today's ATE base, but it shows the pattern: new test sockets tend to form as duopolies too.

What would change this read is a third supplier qualifying at leading-edge compute test, or Chinese vendors moving from mature node into AI-adjacent work. Neither has happened. The evidence so far suggests the constraint is not engineering alone; it is whether customers will fund and run a third qualification, and the public disclosure does not settle that.

02

Competitive advantage

Does the company have a moat? Fairly strong, in the smaller half of a two-firm market

Moat: Fairly Strong. Teradyne and Advantest collect roughly 85% of what chipmakers spend on automated test equipment, and Teradyne has cleared its cost of capital in each of the past five fiscal years, including the 2023-2024 downturn. Persistent concentration plus excess returns that survive a bad year is the signature of entry barriers. Two facts keep the rating short of Strong: Teradyne is the smaller of the two vendors, and its share has sat near 30% instead of climbing.

Profits on their own prove little, because a well-run firm can out-execute rivals for a stretch and, if competitors copy what works, the returns get competed away. What separates a structural barrier from good management is that the market structure and the profitability both hold. Test equipment passes that test. The leading-edge supplier list stays short, and the weakest of the past five years still produced an 18% return on equity.

Market position: Moderate, with a share that has stopped moving

Teradyne put its own ATE share at about 30% at Bank of America's Global Technology Conference in June 2026. Advantest holds something close to half, and the rest sits with Cohu, Chroma, HYC and a group of Chinese suppliers confined largely to mature-node test. That makes Teradyne the clear No. 2 in a two-firm market.

A 30% share and second place fall short of what a Strong position rating requires. The stability test cuts the other way. Greenwald's rule of thumb is that a five-year share swing beyond five percentage points points to absent barriers, while a swing under two points points to strong ones. The test cannot be run cleanly here: neither Teradyne nor Advantest publishes a year-by-year ATE share series, so the five-year table that would settle it does not exist in public disclosure. What is on record is narrower. Management said dual-vendor strategies at major customers hold Teradyne's share near 30% in the near term, with a target of 35% to 38% over three to four years, and said the company holds strong share in network-switch test, where it expects a two-vendor structure to form as co-packaged optics ramps. Share stability: Fairly Strong, on the strength of a duopoly that has held together, tempered by the missing annual series.

Profitability: Strong, and it held through the last downturn

Returns have been high enough and consistent enough to clear the framework's bar. The table uses return on equity because a clean multiyear ROIC series is not disclosed; a computed ROIC follows underneath as a cross-check.

Fiscal yearRevenue (USD bn)Gross marginNet marginROE
20213.7059.7%27.6%42.5%
20223.1659.2%22.8%28.5%
20232.6857.5%16.8%18.0%
20242.8258.5%19.2%20.3%
20253.1958.6%17.2%19.6%

*Computed from FY2025 net income of $550 million and year-end equity of $2.80 billion.

The five-year average ROE is about 26%. Revenue fell 28% from $3.70 billion in FY2021 to $2.68 billion in FY2023 and net income fell 56%, yet ROE never dropped below 18% and gross margin stayed inside a 57%-60% band. A rough ROIC, taking after-tax EBIT over equity plus debt less cash, runs from about 70% in FY2021 to roughly 20% in FY2025, with the two years in between near 24%. That calculation is an estimate from reported figures, and the treatment of cash and goodwill moves it, but the direction is not in doubt: the worst year of the cycle still cleared the 15% floor that the framework associates with a durable advantage. The window here is five years, not the ten the benchmark calls for, and it contains one boom and one bust.

Peer margin comparisons are hard to make on like-for-like terms because Advantest reports under different conventions, but the durability of Teradyne's own gross margin through a 28% revenue decline says more than a single-year comparison would. Pricing did not collapse when volumes did.

Overall: Fairly Strong

Rated Fairly Strong. A two-supplier market with an 85% combined share and returns in the high teens or better through a full cycle is what protected economics look like. The deductions are specific: Teradyne is the smaller vendor, its share has been flat near 30% rather than expanding, and the fastest-growing socket in AI compute test sits with Advantest.

Where the advantage comes from and how long it can last: qualification lock-in and a two-firm R&D race

Two mechanisms carry the weight. The first is the switching cost that wraps around a qualified test socket. The second is the fixed cost of staying on the leading edge of test instrumentation when only two vendors can fund it. Both are real, and neither is absolute.

Customer captivity: Moderate, because customers buy two

A tester gets chosen before a chip reaches volume production. The customer's engineering team writes the test program against that architecture, validates it against known-good and known-bad parts, then runs it for the life of the program. Moving to a different tester means rewriting and revalidating the program, requalifying the handler and the interface hardware, and accepting the risk that a bad part escapes into a system where the cost of failure dwarfs the price of the tool. Those costs are real, and they generate follow-on revenue: service, spares, software and field applications support on an installed base that persists for years after the tool ships.

Two things cap the effect. Customers refuse to be single-sourced, because one supplier hands away price and supply leverage, so any given program is split and Teradyne's realistic ceiling on it is a portion. More importantly, the strongest current sockets are not Teradyne's. Morningstar's analyst writes that Advantest holds the superior position in GPU testing and that head-to-head competition with Advantest is the main drag on Teradyne's growth and profitability. Switching costs protect whoever is already installed, and on the largest growth socket that is the rival.

Rate the source Moderate. A chipmaker that has run a program on Teradyne hardware has a reason to keep the next design there, but nothing prevents it from moving, and the dual-source convention means even a fully committed customer is only half a customer. This is captivity negotiated rather than imposed.

Economies of scale: Fairly Strong against everyone except the leader

Test equipment is a fixed-cost business. R&D is the largest of those costs: instrumentation, custom high-speed ASICs, the software that runs the test program, and a global field organization to install and support it. Spread across a 30% share of a market the company sizes at $12 billion to $14 billion, that fixed cost per dollar of revenue is a fraction of what a vendor with low single-digit share carries. A challenger at Cohu's or Chroma's scale cannot fund the next node's roadmap and still price below the leaders, which is why the field at the leading edge never widens.

The installed base reinforces the same mechanism. Thousands of systems in the field require spare parts inventory, calibration capacity and field application engineers in every region where the customers run fabs. That infrastructure is fixed, and a new entrant would have to build it before winning its first high-volume socket, which means years of spending ahead of revenue.

The critical combination is scale plus customer captivity, because it forces a challenger to out-spend on R&D and dislodge installed programs at the same time. Teradyne has that combination against Cohu, Chroma, HYC and the Chinese suppliers working mature nodes. It does not have it against Advantest, which is larger and therefore carries the lower fixed cost per revenue dollar. So the scale edge protects the No. 2 slot and the economics attached to it. It does not open a path to leadership. Rate this source Fairly Strong, with the caveat that the relative-share advantage runs the wrong way against the one competitor that matters most.

Durability: the risk sits in test intensity, not in a rival brand

The clearest durability risk is cyclical, not competitive. Revenue fell 28% between FY2021 and FY2023 while returns held, which tells you the moat protects profitability, not the top line. Current demand is concentrated in AI: Teradyne reported third-quarter 2026 revenue of $1.31 billion, up 70% year over year, with AI contributing about 60% of sales, gross margin of 59.1%, and memory test revenue of $301 million, a fourth consecutive quarterly record on HBM and DRAM demand. If that capex wave pauses, revenue would fall hard again from a much higher base than in 2023, and the return profile would be tested at a scale it has not yet seen.

The substitute question is less about a competing tester brand than about who does the testing. Built-in self-test and design-for-test move verification into the chip itself and can shrink the number of external insertions, and wafer-level test shifts insertions earlier in the flow. So far complexity has pushed the other way: HBM stacks and advanced packaging add insertions, which is why memory test is setting records. The number of test insertions per device decides whether the moat's value grows or shrinks, and no public series tracks it. That is the variable to watch, and the direction is genuinely unsettled.

A third leading-edge supplier would change the structure more directly. Chinese vendors currently work mature nodes; moving them into AI-adjacent test would break the duopoly, and customers' second-source habit gives them a reason to fund the attempt. Teradyne's own projections put co-packaged optics test at $100 million in 2026, $300 million to $700 million by 2028 and above $1 billion by 2030, on an assumed two-vendor structure. Those are company forecasts tied to a capex path, not reported results, and they are the kind of adjacent socket where a second supplier tends to emerge rather than stay away.

What would move the assessment: Teradyne converting the 30% share into management's 35%-38% target would show the captivity is stronger than the dual-source convention suggests. Losing a flagship compute socket, or a credible third qualification at the leading edge, would show the opposite. So far the evidence shows a protected franchise earning strong returns, with a ceiling on that protection set as much by customer policy as by competition.

Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.

Continue the comparison

Continue reading the full report

  1. Comparison and summaryHow Teradyne Inc compares with its closest rivals, and our moat rating

Free to sign up · 20 research points · unlocking this report uses 10

20 points cover 2 full reports. No card needed.

Research reflects its generation date. Preview the first two modules.

More on Teradyne Inc