Besi Says Hybrid Bonding Orders Are Already 2x All of 2025 and Dismisses ASML as a Threat, but Flat Q4 Guide Tests the Multiple
Q3 2026 earnings call, October 22, 2026: record profitability, EUR 100 million buyback, and a flat revenue outlook
BE Semiconductor Industries reported third quarter revenue of EUR 283.4 million, up 13.4% sequentially and 113.6% year over year, with orders of EUR 342.6 million, up 17% quarter over quarter and 96.1% year over year. Net income reached EUR 107.2 million, a 37.8% net margin that is double the level of a year ago. The headline that matters for the stock is less the beat than the guide: CEO Richard Blickman expects Q4 revenue to be "flat plus/minus 5%," even as orders remain strong. Management insists the flat guide reflects delivery scheduling rather than any softening in demand, and that point drew the most scrutiny on the call.
Hybrid bonding: orders already double all of 2025, with 23 customers and more than 215 systems
Year-to-date hybrid bonding orders have more than doubled versus full-year 2025, driven by HBM programs, more logic chiplet programs entering production, faster-than-expected co-packaged optics adoption, and the start of edge AI device production. Customer adoption has reached 23 customers, and cumulative unit demand has exceeded 215 systems. Besi received Q3 hybrid bonding orders across logic, co-packaged optics and memory.
The most notable disclosure was an order for both hybrid bonding and TC Next systems from a leading memory customer. Blickman said the two tools serve separate product road maps at that customer, and that the TC order was a repeat. He declined to name the customer, but the logic that Needham pressed, that all three HBM suppliers already own Besi hybrid bonders, was not disputed. In Blickman's framing, "why would more hybrid bonders be needed to test, to qualify preproduction if there is not a high-volume production somewhere inside."
Besi was candid that near-term HBM adoption has been slower than hoped. Severe memory shortages this year have extended the use of lower stack heights in current HBM production beyond prior expectations, which pushes back adoption of the superior technology. Management does not see this as a showstopper: "That only relates to satisfying current demand in current end applications," Blickman said of the despeccing noise, adding that road map work continues "at full speed with all 3 memory manufacturers." He argued the adoption decision is not about 12, 16 or 20-high stacks but about energy, latency, bandwidth and heat, citing disclosed performance improvements of 20% to 30% and lower temperatures, and said a large end customer will eventually dictate "I want devices to be hybrid bonded because of better performance." Besi expects hybrid bonding across a range of stack heights over the next 2 years, and direct DRAM-on-accelerator bonding by the end of the decade. Management would not give timing for volume, saying that is determined by the end customers. Investors should note that this leaves HBM as a visible but undated option rather than a 2027 revenue driver.
On logic, Blickman said the leading U.S. logic customer ordered a large volume in 2024, shipped in 2025, and is now preparing for mass production. With that customer's recovering fortunes, he expects further expansion next year, including plans for Asian operations close to Besi's manufacturing. On the leading GPU maker, the timeline analysts recall, qualification this year, orders next year and volume in 2028, remains "more or less" intact, but he said sizing the opportunity would be "a guess."
ASML: management rejects the competitive threat narrative
Blickman took the unusual step of addressing recent analyst reports directly: "We do not view ASML as a competitor, contrary to recent analyst reports." His understanding is that ASML is focused on a decade-long road map for 3D transistor stacking inside the chip, once hybrid bonding becomes a mainstream requirement, and is not competing in die-to-wafer bonding. He cited the size and current state of the market as reasons ASML is uninterested. The two companies are jointly running die stress programs, since stress introduced during layering and dicing determines placement accuracy and throughput, with work taking place inside both companies and at other European sites.
Blickman also used the question to lay out Besi's accuracy road map. Roughly 100 nanometer placement is the standard, and all 215 systems sold are Generation 1 or 1 plus. Besi is shipping its first 50 nanometer Generation 2 systems this quarter, which he said will be sufficient for the next 4 or 5 years. A 25 nanometer system is under development, with a further step to 10 nanometers beyond that, all still chip-to-wafer. He dismissed the idea that ASML would take over Besi's position "in the next couple of weeks" as a "quick conclusion." The answer was reasoned, but it did not address the IP question directly, and some investors will remain unconvinced about longer-term exposure.
The flat Q4 guide: scheduling, not softening
Revenue guidance of flat plus or minus 5% was framed in the release as customers requesting delivery dates aligned with first-half capacity plans. Morgan Stanley's Lee Simpson read this as a customer pushout, and Blickman corrected him: "We have presented this message incorrectly. It's not a customer pushout." Orders of more than EUR 340 million do not all ship immediately; some ship in Q4, some in Q1 and some in Q2, depending on the readiness of the full line and facility infrastructure. Visibility has lengthened from roughly one to 1.5 quarters historically to two quarters or six months, and longer for some programs.
Gross margin is guided to 63% to 65%, the same range as Q3, with Blickman stating there are no one-offs and that mix drives the outcome. Operating expenses are guided up 5% to 10% sequentially, primarily because of strategic consulting costs, which he described as a one-off. Baseline operating expenses fell to 16.4% of revenue from 32% a year ago, the main source of the operating leverage.
Order momentum, photonics, China and power
Blickman said order momentum is expected to continue in Q4 across both traditional and hybrid systems, and that it has continued "until October 22." He framed the key risk clearly: a slowdown in orders would raise the question of "an overall overcapacity" versus a timing issue. Management sees no such signal yet. Demand remains strong for pluggable optics tools tied to data centers, and for the co-packaged optics programs that drew a substantial order in Q1 and several in Q3. Management could not say whether CPO demand is for switches or GPU optical I/O.
China reached 40% of orders and drove the majority of the sequential increase, according to an analyst on the call. Blickman said the demand is broad, spanning high-end smartphone supply chains such as Foxconn, pluggables, TC and hybrid bonding, and he expects it to have "a long-term future." He did not directly address the analyst's concern about capacity built ahead of end demand.
Power management is a growing contributor. Automotive once represented 15% to 20% of revenue and has fallen to roughly 10% as AI grew, but power devices for AI data center modules, built on copper frames with soldered power devices, are restoring that share. A modest recovery in automotive, industrial and non-AI computing is also visible, and Blickman expects the cycle to continue into 2027 if historical patterns hold.
TC Next and the Applied Materials partnership
TC Next now has 7 customers, most of them ordering for logic. The tool targets bond pad pitches below 20 microns and fluxless processing, with Imec data showing capability below 10 microns, "very close to hybrid bonding." Blickman positioned it as an option for customers to choose reflow or hybrid bonding. He said the segment "could be a very sizable product range" over the next couple of years, though he did not give revenue timing.
Besi has joined Applied Materials' EPIC Center in Sunnyvale as an innovation partner, with the first program covering hybrid bonding and TC for a major U.S. customer. Blickman said the collaboration is aimed at bonding processes, materials and cleaning, not at jointly developed integrated systems of the Kinex type. Development also continues at Imec, in Radfeld and in Singapore.
Market backdrop, capital return and strategic review
Besi cited TechInsights' upgraded 2026 assembly equipment forecast of 37% growth, almost double the prior estimate, with the market reaching $12 billion by 2028, and said it expects to grow at more than twice the underlying market rate. Net cash reached EUR 246.6 million, and the buyback increases to EUR 100 million for the annual period starting October 27.
A 16-week strategic review with key customers and Applied Materials is under way, 4 weeks in, with conclusions expected in early January. Blickman said it will focus on how hybrid bonding, TC and mass reflow road maps determine demand over the next 3 to 5 years, and what is missing in advanced packaging. The previous review, in early 2025, led to higher hybrid bonding market targets, so investors should expect updated long-term targets in January.