BE Semiconductor Industries N.V. (BESI.AS) - Deep Dive Research Report

Euronext Amsterdam: BESI | US OTC/ADR: BESIY | Sector: Technology (Semiconductor Assembly Equipment) | Report date: 24 July 2026

1. What the company does

Besi builds the machines that put finished computer chips together. It does not make chips. It makes the capital equipment that sits at the very back end of a semiconductor factory, in the step called "assembly" or "packaging," where a bare silicon die that has just come off a wafer is picked up, precisely positioned, and permanently joined to another die, to a substrate, or into a stack. If you have heard that the hardest problem in AI hardware right now is not the transistor but how you glue a dozen chiplets and stacks of high-bandwidth memory into one package, Besi sells the tools that do the gluing.

In plainer terms: a modern AI accelerator like an Nvidia GPU is not one chip. It is a logic die, several stacks of HBM (High Bandwidth Memory), and an interposer, all mounted next to and on top of each other with connections finer than a human hair. The machine that places those dies, aligns them to sub-micron accuracy, and bonds them is a "die bonder." Besi is one of three companies on earth that matters in high-end die bonding, and it is the technology leader in the newest and most demanding version of that step, called hybrid bonding.

The company was assembled, appropriately enough, by bonding pieces together. Besi was formed in 1995 around a small US die-bonding operation and floated on Nasdaq and Amsterdam that year under Richard Blickman, who still runs it three decades later. Rather than grow one product organically, Besi spent twenty years buying the specialist European and Asian toolmakers whose brands it still sells under: Meco (plating, Netherlands), Fico (molding, Netherlands), Datacon (flip-chip and advanced die attach, Austria), and Esec (die bonding, Switzerland). It bolted these acquisitions onto a common operating model: keep the process R&D in high-cost Europe, move the manufacturing and assembly of the machines themselves to low-cost Asia (Malaysia, China, and now Vietnam). That structural decision - European engineering, Asian production - is why Besi has consistently run gross margins in the low-to-mid 60s percent and among the highest operating margins in all of semiconductor equipment.

The core value proposition is accuracy at speed. Placing a die is trivial if you have all day; the entire game is placing it to within a few tens of nanometers of alignment, thousands of times per hour, without a single failure, because a misplaced die in a $30,000 AI package ruins the whole unit. Besi's fifteen-plus years of process knowledge in ultra-high-accuracy placement, thermal management during bonding, and machine vision is the moat. A new entrant cannot buy that; it has to be earned die by die, customer qualification by customer qualification.

CEO Richard Blickman's framing on the Q2-26 call (23 July 2026): growth was "primarily due to significantly expanded AI infrastructure spending, a modest recovery in traditional mobile and auto/industrial markets and disciplined overhead management."

A concrete walk-through: a leading logic foundry (read: TSMC-class) wants to build a next-generation AI processor where two logic dies are joined face-to-face with copper-to-copper connections and no solder bumps at all - hybrid bonding. It qualifies Besi's Datacon 8800 CHAMEO hybrid bonder over a two-to-three-year evaluation. Besi ships evaluation tools, the customer runs yield trials, and once qualified, orders production lines. Besi has disclosed that one leading logic customer installed six integrated hybrid-bonding production lines incorporating 30 Besi bonders. Each line is a multi-year, multi-tens-of-millions-of-euro commitment, and once a fab is qualified on Besi's tool, switching to a competitor means re-qualifying an entire process - which is why this business, once won, tends to stay won.

2. Business segments

Besi reports under three product groups. The split is heavily weighted toward one of them, but understanding all three matters because they share a common assembly-line context and because the smaller two are the cash-cow and the option, respectively.

Die Attach (~80% of 2025 revenue)

This is the business. Die Attach covers every machine whose job is to pick a die and bond it somewhere: mainstream die bonding (Esec-brand systems that attach chips to leadframes and substrates for automotive, industrial, mobile), flip-chip bonders (Datacon FC QUANTUM), multi-module attach (Datacon 2200 evo), thermocompression bonding or TCB (the 9800 TC and "TC Next" lines), and the crown jewel, hybrid bonding (Datacon 8800 CHAMEO).

The core capability is placement accuracy under thermal and mechanical stress. Flip-chip and multi-module systems have, in management's words, "gained significant share in the market for AI-related 2.5D assembly structures" - that is, the CoWoS-style packages where logic and HBM sit side by side on an interposer. TCB is the incumbent method for stacking HBM today; hybrid bonding is the method the industry is migrating toward for the densest interconnects, because it eliminates solder bumps entirely and allows far finer pitch. Besi's hybrid-bonding roadmap targets 25-nanometer placement accuracy and throughput above 5,000 units per hour, and it had over 150 hybrid bonders installed by end-2025.

This segment is simultaneously the margin engine and the growth bet. Its mainstream die-bonding portion is cyclical and moves with mobile and automotive demand; its advanced portion (flip chip, TCB, hybrid bonding) is the structural AI-driven growth story. When management talks about the future, they are almost always talking about this segment.

Packaging (~17% of 2025 revenue)

Sold under the Fico brand, this group makes the equipment for the steps after the die is bonded: molding systems (AMS-X, MMS-X) that encapsulate the assembled package in protective compound, trim-and-form equipment (Compact Line) that shapes the leadframe leads, and singulation/sawing systems that cut the molded strip into individual finished packages.

The core capability here is high-throughput precision handling of fragile molded strips at scale - less exotic than hybrid bonding, but a mature, defensible niche where Besi and a handful of competitors have consolidated the market. It exists as a separate group largely for acquisition-history reasons (Fico was a distinct Dutch company) and because molding and singulation are chemically and mechanically different disciplines from die placement. Strategically this is the steady contributor: it rides the same package volumes as Die Attach but with less of the advanced-packaging upside and less cyclicality at the extremes.

Plating (~3% of 2025 revenue)

The Meco brand makes continuous plating and leadframe-cleaning systems: electrochemical deposition of metal onto leadframes and connectors, plus adjacent niches in connector plating and solar plating. It is the smallest and oldest of the three, a specialist wet-chemistry business bolted on decades ago. Its core capability is high-volume electroplating uniformity. Strategically it is a cash cow and a rounding error on the growth story, retained because it is profitable and shares the same semiconductor-assembly customer base. Management rarely dwells on it.

3. Products and business detail

Die Attach catalogue. The Esec-branded systems (e.g. Esec 2009 / 2100 soft-solder die bonders) are the mainstream workhorses - high-speed epoxy and soft-solder die attach for automotive, LED, discrete and mainstream logic. The Datacon-branded systems climb the accuracy ladder: FC QUANTUM for flip chip, the 2200 evo family for multi-module and 2.5D assembly, the 9800 TC and "TC Next" for thermocompression bonding of stacked memory and fine-pitch chiplets, and the 8800 CHAMEO for hybrid bonding. The distinction that matters: TCB uses heat and pressure to reflow micro-bumps between stacked dies; hybrid bonding eliminates bumps altogether, joining copper pads directly to copper in a dielectric-to-dielectric bond. Hybrid bonding is harder (it demands extreme surface cleanliness and 25-50nm alignment) but delivers far higher interconnect density, which is why it is the technology the whole industry is qualifying for the 2027-2030 product cycle.

Packaging catalogue. Fico AMS-X and MMS-X molding systems encapsulate assembled devices; the Compact Line performs trim and form; the Sawing Line singulates. These serve mainstream and automotive packaging houses.

Plating catalogue. Meco continuous plating lines and leadframe cleaning systems, plus connector and solar plating niches.

The manufacturing model. This is the operational heart of Besi. Process development and prototype engineering sit in Europe - the Netherlands (Duiven, headquarters and Meco/Fico heritage), Austria (Radfeld, the Datacon advanced die-attach centre), and Switzerland (Cham, the Esec die-bonding centre). Volume manufacturing and machine assembly are in Asia - Malaysia and China - with new capacity being stood up in Vietnam. On the Q1-26 call management said it was "establishing additional Vietnamese production capacity for mainstream assembly to free up Malaysian facilities for wafer-level work," a telling move: shift the low-mix mainstream machine build to the cheapest location so the higher-value hybrid-bonding and wafer-level capacity in Malaysia can expand.

Capacity as a growth lever. Because hybrid bonders are the constrained product, Besi has been explicit about scaling: on the Q1-26 call it put current hybrid-bonder capacity at roughly 15 units per month (~180 per year), expandable to about 250 per year, with standardized six-month lead times. Building the machines is itself a bottleneck the company is deliberately widening ahead of the memory-adoption wave.

Milestones that shaped the business. The Datacon (2005) and Esec (2009) acquisitions gave Besi its advanced die-attach and die-bonding franchises. The 2020 start of collaboration with Applied Materials on die-based hybrid bonding - and Applied's subsequent ~9% equity stake announced in 2025 - locked in a partnership with the world's largest wafer-fab-equipment maker on the technology that now drives the whole thesis. The June 2025 Investor Day, where management raised its long-term revenue target to €1.7-2.2bn and EBIT-margin target to 45-55%, marked the point at which Besi formally re-based its ambitions around AI packaging.

4. Customers

Besi sells to two overlapping customer types: the leading logic foundries and IDMs (integrated device manufacturers) at the frontier, and the large Asian OSATs (outsourced assembly and test subcontractors) that do packaging at volume. Named central players in its end-market ecosystem include TSMC-class logic foundries, the big memory makers (SK Hynix, Samsung, Micron) for HBM, and the OSAT giants (ASE, Amkor, and Chinese subcontractors). Geographically, the customer base is overwhelmingly Asian: in Q2-26, 39% of revenue was China, 49% Asia-Pacific ex-China, and only 12% Europe/US/other.

Inside these customers, the buying decision for advanced tools sits with the advanced-packaging process-engineering and technology-development groups, not a procurement desk. The criteria are yield, placement accuracy, throughput, and cost-of-ownership, proven over a long qualification. The sales cycle for a frontier tool like a hybrid bonder is measured in years: Besi ships evaluation tools, the customer runs qualification lots, and only after the process is locked do production orders follow. Management has repeatedly framed hybrid-bonding adoption as a customer-count metric precisely because each new qualified customer is a hard-won, durable relationship - adoption went from 15 customers at end-2025 to 21 by end-Q2-26.

Why they choose Besi: at the top of the accuracy curve, there are only a few credible suppliers, and Besi (partnered with Applied Materials) is the technology leader in hybrid bonding. Switching costs are severe. Qualifying a die bonder into a high-volume advanced-packaging line takes months to years of yield validation; once a fab standardizes on a tool and trains its process around it, moving to a rival means re-qualifying the entire flow and risking yield on a product worth tens of thousands of euros per package. That installed-base lock-in is the reason Besi's advanced franchise, once won, compounds.

Concentration is real and is a feature of the industry, not a Besi-specific flaw: the frontier of advanced packaging is a handful of foundries and memory makers, so a small number of large accounts can drive order swings quarter to quarter. Contract structure is largely order-based capital equipment (not recurring subscription), which is why Besi's revenue is inherently lumpy and cyclical - orders can more than double year-over-year (H1-26 orders +116.5%) or fall a third (Q2-25 orders -30.9%) depending on where customers sit in their capex cycle.

5. Competitive landscape

Besi competes in back-end assembly equipment, a market dominated by three scaled players plus a cluster of specialists in the specific niche of hybrid bonding.

The three-way core of high-end die bonding is Besi, ASMPT, and Kulicke & Soffa. ASMPT is the largest back-end assembly-and-packaging equipment vendor by revenue, spanning die bonders, wire bonders and surface-mount, and it uses that scale and breadth to pressure rivals on speed-to-market. Kulicke & Soffa is the historic wire-bonding leader pushing into advanced packaging and TCB. Besi wins at the top of the accuracy curve - flip chip, TCB Next, and especially hybrid bonding - where its process knowledge and Applied Materials partnership give it the technology edge; it is more exposed than ASMPT in mainstream breadth, where ASMPT's scale is hard to match.

In hybrid bonding specifically, the field widens to include EV Group and SUSS MicroTec (wafer-to-wafer bonding specialists), Applied Materials and Tokyo Electron (adjacent front-end giants moving into the space, with Applied a Besi partner and shareholder), and Hanmi Semiconductor, which has grown fast in thermocompression bonding for HBM memory customers. Industry commentary puts EV Group, SUSS and the die-to-wafer players collectively at roughly 60-65% of the hybrid-bonding market when wafer-level is included; Besi's stronghold is die-to-wafer and die-to-die hybrid bonding, the flavor most relevant to logic and HBM stacking.

The barriers to entry are high but not absolute. They are built from process know-how (fifteen-plus years of placement-accuracy engineering), customer qualification lock-in, and the installed base. What keeps the barrier from being a fortress is that the adjacent front-end giants (Applied, TEL) have deep pockets and the memory-focused specialists (Hanmi) are advancing quickly in TCB. The structural shift underway is the migration from bumped TCB toward bumpless hybrid bonding for the densest packages - a shift that favors Besi if it holds its technology lead, but where the memory makers' eventual technology choice (TCB vs hybrid bonding for HBM stacking) is not yet settled and will reshape the competitive map.

Market caps are peer-size references only, drawn from mid-2026 data, and move daily.

6. Industry

Demand for Besi's equipment is driven by one thing above all others right now: the physical limits of chip scaling and the rise of AI compute. As transistor shrinks slow, the industry has shifted the burden of performance gains onto packaging - stacking dies, placing chiplets side by side, and connecting them with ever-finer interconnects. Every AI accelerator, every HBM stack, every co-packaged optics module needs more advanced assembly steps than the chip it replaces, and those steps require Besi's tools. Roughly 70% of Besi's 2025 revenue was for advanced-packaging applications, and AI-driven computing exceeded 50% of the business mix in 2025, overtaking mobile; by H1-26, AI systems were about 60% of orders.

On size and trajectory: third-party and management framing point to the assembly-equipment market growing on the order of 70%+ from 2025 to 2030, with advanced packaging outpacing mainstream. The narrower die-bonder equipment market is put at roughly US$1bn in 2025. The hybrid-bonding equipment segment is small today (industry estimates around US$150-170m in 2025) but is forecast to grow at 20%+ CAGR into the 2030s; Besi's own Investor Day framing sees hybrid bonding plus advanced TCB reaching around €1.65bn by 2030, and hybrid bonding becoming the single largest assembly segment. These are cited estimates from Besi's Investor Day and third-party research houses (Mordor, InsightAce); the exact numbers vary by source, but the direction is consistent.

Besi sits high in the value chain: it is a critical-path equipment supplier to the foundries, memory makers and OSATs that assemble AI silicon. It is not a commodity toolmaker; at the frontier of accuracy it is one of a very small number of qualified suppliers, which is why its margins are among the best in equipment.

Regulation and geopolitics matter through two channels: export controls (advanced-packaging tools sit near the line of US/allied restrictions on China's access to leading-edge capability) and tariffs (management flagged tariff uncertainty as a swing factor for mainstream demand through 2025). With 39% of Q2-26 revenue in China, any tightening of controls on advanced-packaging equipment sales to Chinese customers is a live risk.

Cyclicality is the industry's defining trait. Semiconductor capital equipment moves in violent cycles tied to customer capex. Besi lived this in real time: revenue actually fell 2.7% in 2025 as mainstream mobile and auto stayed weak, then orders surged 116% in the first half of 2026 as AI capex accelerated. The tailwind is the secular AI/advanced-packaging build-out; the headwind is that this is a capex business where a single quarter's orders can swing 30-40% and the mainstream end markets (mobile, auto, industrial) remain cyclical and, for now, only modestly recovering.

7. Growth triggers

Every trigger below is drawn from the six most recent concalls, cited by quarter.

• Hybrid-bonding customer adoption ramping toward the 2027-2030 product cycle. Adoption climbed from ~50 systems / a handful of customers (Q1-25) to 18 customers and 150+ systems (Q4-25) to 20 customers (Q1-26) to 21 customers by end-Q2-26. Repeated across all six calls - the central growth engine.

  "Hybrid bonding adoption is increasing as the 2027-2030 product cycle approaches, with new applications announced in logic, memory, co-packaged optics, and consumer segments." (Q1-26 concall, Apr 2026)

• HBM memory adoption of hybrid bonding - the swing catalyst. Besi shipped two evaluation tools to a second memory customer for HBM applications (Q1-26), and management framed 2026 as decisive for whether HBM stacking moves to hybrid bonding. (Q4-25 concall, Feb 2026; Q1-26 concall, Apr 2026)

  "2026 will be a very important year to understand the adoption of hybrid bonding for HBM stacking." (Q4-25 concall, Feb 2026)

• Integrated hybrid-bonding production lines at a leading logic customer. Six integrated lines incorporating 30 Besi bonders installed at one leading logic customer (Q4-25 concall, Feb 2026) - evidence the technology is moving from evaluation to volume production.

• AI-related 2.5D assembly share gains via flip-chip and multi-module systems. Besi said these systems "gained significant share" in AI-related 2.5D data-center assembly; die-attach bookings from Asian subcontractors for 2.5D datacenter drove the order recovery. (Q3-25 concall, Oct 2025; Q4-25 concall, Feb 2026)

• Photonics / co-packaged optics as a new demand leg. Renewed capacity purchases by leading photonics customers were a named order driver (Q3-25), and photonics/CPO recurred as a strength through H1-26. (Q3-25 concall; Q2-26 concall, Jul 2026)

• Capacity expansion ahead of demand. Hybrid-bonder capacity being scaled from ~15/month (~180/year) toward ~250/year; new Vietnam production capacity for mainstream assembly to free Malaysian facilities for wafer-level work. (Q1-26 concall, Apr 2026)

• TC Next adoption broadening. TC Next expanded to five customers across logic, memory and photonics. (Q4-25 concall, Feb 2026)

• Raised long-term targets at the June 2025 Investor Day. Revenue target lifted to €1.7-2.2bn and EBIT-margin target to 45-55%, reflecting confidence in AI packaging demand. (Referenced in Q2-25/Q3-25 calls; Investor Day 12 Jun 2025)

• Sequential revenue growth continuing into H2-26. Q3-26 guided to +10-15% versus Q2-26, on top of Q2-26's +35% sequential jump. (Q2-26 concall, Jul 2026)

8. Key risks

HBM technology-choice risk - the single biggest binary. Besi's bull case leans heavily on high-bandwidth memory migrating from thermocompression bonding to hybrid bonding. If the memory makers (SK Hynix, Samsung, Micron) decide TCB is good enough for the next few HBM generations, or standardize on a rival's TCB tool, a large chunk of Besi's addressable growth slips years to the right. Management has been unusually honest that this is unresolved - flagging 2026 as the year the answer emerges and noting Samsung was expected to clarify its roadmap by Q2-26. This is a moderate-probability, high-impact risk sitting at the heart of the thesis.

Extreme order cyclicality. This is a capital-equipment business with order-based, lumpy revenue. Orders fell 30.9% year-over-year in Q2-25 and then rose 128.8% year-over-year in Q2-26. A reader cannot extrapolate any single quarter. A pause in AI capex, an inventory correction at the OSATs, or a delayed customer program can turn a booming order book into a flat one within two quarters. High probability of volatility; the question is only amplitude.

Customer concentration at the frontier. The advanced-packaging frontier is a handful of foundries and memory makers. A single large customer delaying or pulling a production ramp swings quarterly orders materially. This is inherent to serving the leading edge, but it means Besi's results are hostage to a few capex decisions.

China exposure and export controls. With 39% of Q2-26 revenue in China, any tightening of US/allied export controls on advanced-packaging equipment to Chinese customers, or Chinese retaliation, directly hits the top line. Management repeatedly flagged tariff and trade uncertainty as a swing factor for mainstream demand.

Competitive encroachment from deep-pocketed adjacents. Applied Materials and Tokyo Electron have the balance sheets to invest heavily in hybrid bonding, and Hanmi is advancing fast in HBM TCB. Applied is currently a partner and ~9% shareholder, which is protective - but partnerships can shift, and the memory-focused specialists compete directly for the HBM opportunity Besi is chasing.

Margin normalization. Besi's mid-60s gross margins are exceptional and partly reflect a favorable advanced-packaging mix and disciplined overhead. Q3-26 was guided down to 63-65% from 65.7%. A shift back toward mainstream mix, price competition in a downturn, or the cost of ramping capacity ahead of demand could compress margins from today's peak levels.

Mainstream end-market drag. Mobile, automotive and industrial - the bread-and-butter of Die Attach's mainstream portion and of Packaging - were only in "modest recovery" as of Q2-26. If that recovery stalls, the AI story has to carry the entire company, raising the stakes on the HBM binary above.

9. Walk the talk

The six concalls used for this assessment: Q1-25 (Apr 2025), Q2-25 (24 Jul 2025), Q3-25 (Oct 2025), Q4-25/FY-25 (19 Feb 2026), Q1-26 (Apr 2026), and Q2-26 (23 Jul 2026). The most recent is one day old relative to this report.

The story these six calls tell is of a management team that guides near-term numbers with real precision and consistently lands inside its own ranges, while being candid - not promotional - about the one thing it cannot yet control (HBM adoption timing).

Start at Q1-25. Besi actually missed consensus (revenue €144.1m against a ~€160m forecast, EPS €0.40 against ~€0.49) and guided Q2 to "flat ±10%" with gross margin 62-64%. This was the trough of the mainstream cycle, and management did not sugar-coat it - it warned openly about tariff uncertainty and said hybrid-bonding volume production in smartphones and memory was "expected no earlier than 2027." That 2027 framing has been repeated consistently ever since, which is the opposite of a company pulling forward hope to prop up a stock.

Q2-25 delivered inside guidance: revenue €148.1m (+2.8% sequentially, within the flat ±10% band), gross margin 63.3% (inside 62-64%). Management then guided Q3 down 5-15% - and Q3-25 came in down ~10.4%, again squarely inside the range, with operating income at the high end on better-than-forecast margins. Two consecutive quarters of guidance met, including the willingness to guide revenue lower when the cycle demanded it. That candor is the credibility signal.

The inflection is visible from Q3-25 onward. Having guided Q4 to +15-25%, Besi delivered Q4-25 revenue of €166.4m, up 25.4% sequentially - the top of the range - with orders up 43.3%. It then guided Q1-26 to +5-15%; Q1-26 landed at €184.9m, up ~11% sequentially and +28.3% year-over-year, inside the range. Q1-26's guidance for Q2 was the boldest of the set: "30-40% sequential revenue growth" with gross margin 64-66%.

"Revenue will increase between 5% and 15% versus the fourth quarter... with gross margins ranging between 63% and 65%." (Q4-25 concall, Feb 2026) - delivered: Q1-26 revenue +~11% sequentially, gross margin 63.5%.

Q2-26 is where the talk was fully walked. Guided to +30-40% sequentially, Besi delivered +35.2% (revenue €249.9m), with gross margin 65.7%, at the top of the 64-66% range, and net income up 177% year-over-year. The H1-26 guidance given a quarter earlier - "49% revenue increase versus first half of 2025" - landed at +48.8%. That is guidance kept to within a fraction of a point.

On the longer-horizon promise, the June 2025 Investor Day raised the 2030 revenue target to €1.7-2.2bn and EBIT margin to 45-55%. It is too early to score a 2030 target, but the near-term trajectory (H1-26 net margin 32.3%, Q2-26 net margin 35.6%) is running consistent with, not behind, that ambition.

Assessment: this is management that does what it says. Across six quarters spanning a cyclical trough and a sharp AI-led recovery, Besi guided each quarter inside a tight band and hit it every time, including guiding revenue down honestly when the cycle turned against it. On the one item it cannot control - HBM's move to hybrid bonding - it has been disciplined about repeating a 2027 timeline rather than promising it early. The credibility record is strong.

10. Shareholder friendliness index

Dividends. Besi runs a high-payout policy, distributing roughly 95-100% of net income as dividends, so DPS tracks earnings and therefore the cycle. For FY2024 it paid €2.18 per share (about €172.8m, paid in Q2-25). For FY2025 it proposed €1.58 per share (about €125.4m, paid in Q2-26) - a cut, but a mechanical one: 2025 earnings fell in the cyclical trough and the ~95% payout policy passed that through. The exact FY2023 DPS could not be independently verified within this search, but the policy is consistent: Besi pays out nearly all of its earnings, so expect the dividend to rise again as 2026 earnings recover. The high payout ratio is the notable feature - Besi returns essentially all profit rather than hoarding it.

Buybacks and dilution. Besi buys back shares alongside dividends. In H1-25 it deployed €72.2m on buybacks (e.g. €20.7m / 196,000 shares in Q2-25 alone), and across full-year 2025 it distributed €254.8m in combined dividends and buybacks. A current €60m buyback program is running: €40.1m had been deployed by end-Q2-26, including €14.7m (57,000 shares at ~€257.58) in Q2-26 - consistent with the MoatMap-recorded near-daily repurchases of roughly €240,000 per trading day from late May through mid-July 2026. Against this, Besi carries convertible notes that dilute on conversion: in Q2-26 its €175m 2029 convertible notes were converted to equity, issuing about 1.5m treasury shares. Net effect on share count over the period is roughly flat - buybacks are used substantially to offset convertible and option dilution rather than to shrink the count aggressively, so shares outstanding have hovered near ~79-80m rather than declining materially.

Verdict: Returns Capital - Besi pays out nearly all of its earnings as dividends and runs continuous buybacks, though the buybacks mostly offset convertible/option dilution rather than materially retiring stock.

11. Insider activities

Besi is listed on Euronext Amsterdam, so insider dealings are reported under EU Market Abuse Regulation Article 19 (PDMR notifications to the Dutch AFM). The MoatMap disclosure database is used here as the spine for the last 12 months, cross-checked against the venue.

Recent transactions (last 12 months). All recorded insider dealings in the window were open-market sales, clustered in May 2026, by senior operating executives (SVP/VP level) - not by the CEO, CFO, or supervisory board.

Buys - none. There were no open-market insider purchases in the window. The strongest signal in this section (an insider buying with conviction) is simply absent.

Sells - the why. All six sales fall in May 2026, immediately after the stock had run hard on the AI-driven order surge (shares in the €250-290 range versus far lower a year earlier) and around the April AGM / dividend period. The pattern - multiple operating executives (Sales, Packaging, Plating, Asia support) each trimming, with the SVP Packaging's ~€2.8m sale the largest - reads as routine diversification and profit-taking into strength after a large share-price appreciation, not as a signal about the business. None of the sellers is the CEO, CFO or a supervisory director, and the sizes relative to senior-executive compensation are modest (mostly a few months' to about a year's pay). No specific reason (10b5-1-equivalent plan, estate, gift) is disclosed in the filings; where undisclosed, it is left as such rather than guessed. Notably, Berenbak (VP Plating) sold three times - but Plating is the ~3% rump segment, so this is not a read on the core franchise.

Net assessment. Insiders were net sellers over the last 12 months, but the activity is narrow (four operating VPs/SVPs), modest in size, concentrated in a single post-run-up month, and entirely absent of the people whose trades carry the most signal (CEO, CFO, board). This is best read as ordinary diversification into a sharp price rally rather than a negative view on fundamentals - the profile is neutral to mild concern, and it would only escalate if the CEO or CFO began selling or if the selling broadened and persisted. There was no cluster buying and no board-level activity in either direction.

12. Scenarios

Bull case. The HBM binary resolves in Besi's favor. Over the next two to three years, the major memory makers conclude that stacking twelve-plus-high HBM4E and HBM5 dies at the required density is only economic with hybrid bonding, and they qualify Besi's Datacon 8800 CHAMEO into volume production. The 2026 "decisive year" that management flagged turns into a wave of production orders in 2027-2028, on top of the logic-side hybrid-bonding ramp already underway (six integrated lines and 30 bonders at one leading logic customer becomes the template for several). AI infrastructure spending stays elevated, photonics and co-packaged optics add a second growth leg, and the mainstream mobile/auto recovery finally arrives to lift the ~20% of the business that has been dormant. Besi's capacity investments (Vietnam freeing Malaysia for wafer-level, hybrid-bonder output scaling toward 250/year) prove perfectly timed. The €1.7-2.2bn / 45-55% EBIT Investor Day targets look conservative, and Besi consolidates its position as the technology leader of the single most important step in AI packaging, with Applied Materials as a partner rather than a rival.

Base case. Management keeps doing what the six concalls show it doing: guiding tightly and delivering inside the range. AI-driven advanced packaging carries the company, hybrid-bonding customer adoption keeps climbing steadily (it went 15 → 21 in two quarters), and logic customers move from evaluation to production on schedule. HBM adoption of hybrid bonding proceeds but gradually - some memory stays on TCB for another generation or two, so the memory upside is real but slower than the bull case. Mainstream end markets recover modestly rather than sharply. Revenue and margins grind toward the raised long-term targets over the decade, with the usual quarter-to-quarter order lumpiness that makes any single print unreliable. Besi remains a high-margin, high-payout, cyclically-exposed technology leader whose fortunes rise and fall with AI capex but trend structurally upward.

Bear case. The HBM opportunity disappoints. The memory makers standardize on thermocompression bonding for the next several HBM generations, or a rival (Hanmi in TCB, or a deep-pocketed Applied/TEL push in hybrid bonding) wins the memory qualifications, and Besi's biggest single growth lever slips years to the right. Simultaneously, AI capex cools - an inventory correction at the OSATs or a pause in hyperscaler spending turns the 128%-year-over-year order surge into a flat or falling book within a couple of quarters, exactly as the cycle whipsawed from -31% (Q2-25) to +129% (Q2-26). China, at 39% of revenue, gets squeezed by tightened export controls on advanced-packaging equipment. Mainstream mobile and auto stay weak, leaving the AI story to carry a company whose AI story just stalled. Peak mid-60s margins normalize downward as mix shifts and the cost of capacity built ahead of demand bites. Because this is order-based capital equipment with concentrated frontier customers, the downturn would be visible fast and sharp - the same operating leverage that drove net income up 177% in the upcycle works brutally in reverse.

13. Further reading

• Hybrid Bonding Process Flow - Advanced Packaging Part 5 - SemiAnalysis, 2024-02-09 [paid]

Sources: Besi Q2-26/H1-26 results (GlobeNewswire), Besi press releases (IR), Q1-2026 earnings call (Investing.com), Q4-2025 earnings call (Alpha Spread), Q3-2025 earnings call (Alpha Spread), Q2-25 results (GlobeNewswire), Q1-2025 earnings call (Investing.com), Besi products & technology, Besi 2025 Investor Day - increased financial targets, Die Bonder Equipment Market (Mordor Intelligence), Hybrid Bonding Market (InsightAce), ASMPT market cap (companiesmarketcap), Kulicke & Soffa market cap (stockanalysis), Hanmi Semiconductor market cap (stockanalysis), SemiAnalysis - Hybrid Bonding Process Flow