Report Interpretation

Overview

This Bernstein conference-takeaway report on BESI argues that hybrid bonding remains the long-term interconnect technology for logic and memory. Near-term HBM deployment is constrained by process-yield maturation, but management and Bernstein continue to see a substantial adoption ramp from 2028-29, supported by BESI’s established technology position and exposure to advanced packaging.

Core views

Bernstein’s discussions with BESI management reinforced the view that hybrid bonding is becoming the long-term interconnect solution for logic and memory because of its performance, power-efficiency and thermal benefits. The near-term issue in HBM is yield rather than customer interest. Logic applications are already mature: hybrid-bonding yields exceed 99.9% in production programs including AMD, Broadcom, Apple M5-related products and Intel applications. In contrast, HBM yields are generally in the 90% range and can be in the 70% range for more demanding flows. Management attributes the gap to the full process flow—wafer and die preparation, materials and integration—not the bonder alone. Moving from roughly 90% pre-production yields to 99.9% high-volume-manufacturing yields generally takes one to two years. For memory adoption, Samsung is presented as the most vocal potential early adopter, while SK Hynix continues to test hybrid bonding and extend mass reflow. BESI expects Hynix eventually to adopt the technology, with timing as the principal uncertainty. Nvidia is pushing suppliers toward hybrid-bonded HBM for future generations, but BESI expects HBM4E hybrid-bonding volumes to be relatively modest, potentially limited to one layer. The company identifies 2028-29 as the more meaningful volume-ramp period. Even if future HBM specifications are simplified, management believes more complex architectures such as zHBM and multi-layer planar stacking could increase the need for hybrid bonding. The technology roadmap provides a second growth thread. BESI is shipping its first 50nm hybrid-bonding system to TSMC and has internally tested four systems, including systems focused on 50nm accuracy and throughput improvement. The new platform raises throughput from around 1,500 to about 3,000 units per hour, which management says supports a meaningful increase in average selling prices. Qualification may continue for approximately another year, with broader production deployment expected from 2028. BESI retains a 20nm roadmap for 2030-31, while 10nm is described as a mid-to-late-2030s opportunity. CPO, photonics and advanced optical interconnects are additional structural drivers as AI infrastructure scales; management highlighted momentum in pluggables, optical networking and photonics packaging, with Nvidia already deploying hybrid bonding in some applications. BESI’s competitive position rests on more than 180 shipped systems, seven years of process experience and sole-supplier positions at TSMC and Intel for relevant hybrid-bonding applications. TSMC has co-developed the technology with BESI for roughly a decade and, despite evaluating alternatives including Shibaura, has not purchased hybrid-bonding systems from competitors. Management expects some market-share dilution as Korean customers qualify alternatives, but prioritizes process-of-record status, technology leadership and returns over defending a 100% share. Its partnership with Applied Materials is intended to strengthen complete-process capability, process development, customer credibility and qualifications; Applied’s EPIC Centre is expected to open in October with BESI equipment available for integrated process-flow evaluation. The report also stresses that more than 80% of BESI revenue still comes from non-hybrid-bonding products. Traditional packaging demand is supported by the CoWoS ecosystem, alongside photonics, pluggables, connectors, 2.5D and 3D packaging, mobile devices and recovering automotive demand. Fluxless TCB is positioned as a complementary memory technology, with six customers evaluating or deploying systems, while conventional TCB lacks the desired differentiation and margin profile. Dedicated hybrid-bonding cleanroom capacity supports roughly 20-25 systems per month, or 300 annually, and management says capacity can expand with demand. Service revenue is about 15% of sales and could rise toward 20% as front-end-like support needs increase. Management expects hybrid bonding to be margin accretive, with gross margin around 70%, while higher service mix and operating leverage should improve the business profile without a material increase in R&D spending.

Analysis framework

Bernstein summarizes discussions with BESI’s CEO and investor-relations team, then evaluates hybrid-bonding adoption across logic and HBM by comparing yield maturity, customer qualification activity, technology roadmaps, capacity and competitive positioning. It connects these operational factors to revenue mix, margins and a forward P/E-based price target.

Methodology notes

• Industry AnalysisSupply-demand framework

  Technology-adoption and capacity analysis for hybrid bonding

  The report assesses customer adoption, qualification progress, yield constraints and BESI’s production capacity to judge the timing and scale of hybrid-bonding demand.

• Valuation methodsP/E and PEG Valuation

  Forward P/E multiple valuation

  Bernstein sets the €320 target price using a 40x P/E multiple applied to its forward Q5-Q8 EPS estimate.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• BE Semiconductor Industries NV (BESI.NA)

  Primary covered company and supplier of hybrid-bonding and advanced-packaging equipment.

  Strengths

  More than 180 hybrid-bonding systems shipped, seven years of process experience, sole-supplier positions at TSMC and Intel, and a higher-throughput 50nm roadmap.

  Weaknesses

  HBM hybrid-bonding volumes are expected to remain modest near term while customer process yields mature.

  Comparison

  TSMC continues to source hybrid-bonding systems exclusively from BESI despite evaluating alternatives including Shibaura.

  Risks

  Delayed or lower-than-expected hybrid-bonding adoption, market-share loss, weak mainstream assembly recovery, or limited uptake of fluxless TCB.

Key data

• Price target€320.00Bernstein’s target price for BESI.
• Closing price€175.95BESI.NA close on 14 Sep 2026; implied upside is 82%.
• HB logic yieldAbove 99.9%Reported for current production programs.
• HBM hybrid-bonding yieldTypically in the 90% range; sometimes in the 70% rangeLower yields reflect process-flow maturity challenges.
• Hybrid-bonder throughput~1,500 to ~3,000 UPHExpected improvement on the new 50nm platform.
• Hybrid-bonding capacity20-25 systems per month; ~300 annuallyCurrent dedicated cleanroom capacity.
• Service revenue mix~15% of sales, potentially toward ~20%Management sees a larger recurring-revenue contribution over time.
• Hybrid-bonding gross margin~70%Management expects hybrid bonding to be margin accretive.
• Revenue€591.33m F25A; €1,007m F26E; €1,368m F27EBernstein estimates imply a 52.1% CAGR over the displayed period.
• Operating margin29.3% F25A; 41.4% F26E; 45.9% F27EBernstein financial estimates.

Impact & implications

The report argues that HBM-related hybrid-bonding revenue may arrive later than some investors expect because yields need to mature, but this does not alter the longer-term adoption case. BESI’s installed base, sole-source positions, higher-throughput roadmap, service opportunity and premium-margin focus are presented as supports for earnings and margin expansion as adoption broadens.

Risks

• Hybrid-bonding adoption could be lower than expected or delayed in the short term.
• A prolonged recovery delay in the mainstream assembly market could pressure the price-target case.
• BESI could lose hybrid-bonding market share as customers qualify alternative suppliers.
• BESI’s TCB offering may see insufficient customer uptake.

What to watch

• HBM yield progress and the timing of customer qualification toward high-volume manufacturing.
• Samsung, SK Hynix and Micron developments in hybrid bonding and fluxless TCB.
• The scale and timing of HBM4E deployment and the anticipated 2028-29 hybrid-bonding volume ramp.
• TSMC qualification of the 50nm platform and broader production deployment from 2028.
• Demand trends in CPO, photonics, optical interconnects and the CoWoS ecosystem.
• Capacity expansion, service-revenue mix and gross-margin progression as hybrid bonding scales.