Bernstein raises Besi's price target and remains bullish on the logic-chip adoption cycle for hybrid bonding equipment

Overview

This is a Bernstein research report on European and Japanese semiconductor equipment companies, focused on the adoption path of hybrid bonding as a key advanced packaging technology. The report raises its hybrid bonding equipment TAM forecast and increases Besi's price target from €280.00 to €310.00, while maintaining Outperform ratings for Besi and DISCO.

Core views

The report's core view is that logic-chip adoption of die-to-wafer hybrid bonding is clearer than HBM adoption, driven primarily by the TSMC SoIC, AMD, Apple, Google, Nvidia, and Intel Foveros Direct 3D roadmaps. On the HBM side, migration to hybrid bonding may be more gradual because TCB remains effective and there is uncertainty around HBM4E layer counts and height standards, but it should still be supported over the long term by increasing stack heights and expanded I/O channels. Besi, as a leader in hybrid bonding equipment, will benefit directly, while DISCO benefits indirectly through high-cleanliness dicing and grinding equipment.

Analysis framework

The report combines a top-down TAM assessment with adjustments to company earnings forecasts: it first separates logic and HBM applications, estimates 2028 installed bases, unit shipments, customer capacity expansions, and equipment intensity for each, then maps these assumptions to Besi's market share, revenue, EPS, and valuation multiples. For DISCO, it assesses growth leverage from demand drivers including advanced packaging, HBM, CoWoS, NAND CBA, BSPDN, and a recovery in SiC investment.

Methodology notes

• Market sizingTAM segmentation model

  Split hybrid bonding TAM into logic-chip and HBM components

  The report separately estimates 2028 shipments, equipment demand, and revenue for logic D2W hybrid bonding and HBM hybrid bonding, then combines them to derive total TAM of approximately $1.4bn.

• Capacity-driven modelConversion from kwpm to equipment units

  Derive hybrid bonding equipment demand from monthly wafer capacity additions

  The report estimates that each additional 1kwpm of logic SoIC capacity requires approximately 5 to 6 hybrid bonding tools, while HBM requires approximately 2 to 3 tools per 1kwpm because its alignment-accuracy requirements are lower.

• Relative valuationOne-year forward P/E target multiple

  Value forward EPS using a target P/E multiple

  Besi's price target is based on 40x one-year forward P/E and Q5-Q8 EPS estimates. The target multiple is reduced from 45x to 40x to reflect lower hybrid bonding visibility, but the price target rises to €310 due to higher earnings forecasts.

Asset mapping & comparison

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

• BE Semiconductor Industries NV (BESI.NA)

  Core beneficiary of hybrid bonding equipment demand

  Strengths

  The report views Besi as the technology leader in logic hybrid bonding and assumes it will hold approximately 76% combined share in logic and HBM in 2028.

  Weaknesses

  The timing of HBM adoption is uncertain, and Besi may face share competition from Shibaura, ASMPT, TEL, SEMES, Hanwha, Hanmi, and others in both logic and HBM.

  Comparison

  The report's 2028 equipment shipment forecast falls between Besi's base and bull cases, 24% above the base case and 8% below the bull case.

  Risks

  Hybrid bonding adoption below expectations, delayed recovery in the mainstream packaging market, loss of Besi market share, or insufficient adoption of TCB products.

• Disco Corp (6146.JP)

  Indirect beneficiary of advanced packaging and hybrid bonding migration

  Strengths

  Hybrid bonding requires high-cleanliness dicing and grinding systems to ensure particle-free bonding surfaces, benefiting DISCO in the relevant equipment segments.

  Weaknesses

  Although its valuation has declined from its peak, it remains elevated, and growth leverage depends on continued advanced packaging capital expenditure.

  Comparison

  The report says DISCO's one-year forward P/E is 36.5x, down 32% from the recent peak of 53.9x, while remaining bullish on growth drivers.

  Risks

  Excess AI-related GPU and HBM capacity, changes in the competitive environment, and adverse currency movements.

• TSMC SoIC supply chain

  Core driver of logic hybrid bonding demand

  Strengths

  The report expects TSMC SoIC capacity to increase substantially by 2028, with applications from AMD, Apple, Google, Groq, and Nvidia.

  Weaknesses

  The pace of expansion is affected by customer product ramps, advanced packaging yields, and the capital expenditure cycle.

  Comparison

  TSMC and Intel are viewed by the report as leaders in logic hybrid bonding capacity.

  Risks

  Delays to AP7/AP8 expansion, changes in customer chip roadmaps, or advanced packaging demand below expectations.

• HBM equipment chain

  Medium- to long-term incremental scenario for hybrid bonding

  Strengths

  As stack heights and I/O channels increase, HBM has a long-term incentive to migrate from TCB to hybrid bonding.

  Weaknesses

  The report believes TCB may remain mainstream through 2028, while adoption of 16-hi and hybrid bonding for HBM4E remains uncertain.

  Comparison

  Logic adoption has greater visibility than HBM; HBM equipment intensity per kwpm is lower than that for logic.

  Risks

  Looser JEDEC height standards, HBM4E remaining at 12-hi, or improvements in thermal solutions delaying hybrid bonding adoption.

Key data

• 2028 total hybrid bonding TAMapproximately $1.4bnDriven jointly by logic and HBM adoption.
• 2028 logic hybrid bonding TAM$887mnThe report forecasts 237 logic hybrid bonding equipment shipments, a significant increase from the previous forecast of 103 units.
• 2028 HBM hybrid bonding TAM$489mnThe report forecasts 130 HBM hybrid bonding equipment shipments, assuming some HBM4E and HBM5 production migrates to hybrid bonding.
• TSMC SoIC capacity10kwpm in 2025 to 80-85kwpm in 2028The report says TSMC's hybrid bonding capacity expansion is driven by applications from AMD, Apple, Google, Groq, and Nvidia.
• Logic hybrid bonding installed base137 units in 2025 to 582 units in 2028Equivalent to approximately 4.5x growth over three years.
• Besi 2028 hybrid bonding revenueapproximately €901mn or approximately $1.1bnThe report assumes Besi has 80% share in logic and 70% share in HBM, for a combined share of approximately 76%.
• Besi 2028 revenue forecast€1.8bnThe report says this forecast is 11% above market consensus.
• Besi 2028 operating margin48%Driven by operating leverage, with 2028 operating profit forecast at €874mn.
• Besi price target€310.00Previous target €280.00; Outperform maintained.
• DISCO price target¥85,000Outperform maintained.

Impact & implications

If the report's view is realized, hybrid bonding equipment will expand from a niche technology route in high-end packaging into a key equipment cycle for logic chips, compute-memory interconnects, CPO, and part of HBM. Besi has the greatest sensitivity through upward revisions to market share, ASP, and EPS, while DISCO's sensitivity comes mainly from advanced packaging requirements for high-cleanliness dicing and grinding processes. For the industry supply chain, logic adoption has greater visibility than HBM; investment focus should shift from a single HBM narrative toward logic SoIC, chiplets, Foveros Direct 3D, and overall advanced packaging capital expenditure.

Risks

• If HBM4E primarily remains at 12-hi, Samsung's adoption of hybrid bonding could be delayed from 2027 to HBM5 in 2028.
• TCB remains effective for HBM, while looser height standards and improved thermal solutions could delay HBM hybrid bonding penetration.
• Besi may face competition from new entrants in the logic and HBM markets, resulting in a 2028 share below the report's assumption.
• If the mainstream assembly market recovery is delayed until late 2025 or 2026, Besi's traditional business recovery will be constrained.
• Excess AI-related GPU and HBM capacity could lead to DISCO valuation contraction and lower profitability.
• Currency fluctuations could adversely affect DISCO's revenue and earnings.

What to watch

• The actual pace of TSMC SoIC capacity expansion from AP6 to AP7 and AP8, and whether it can reach 80-85kwpm by 2028.
• The timing of volume ramps for logic chiplet applications including AMD, Apple, Google TPU, Groq LPU, and Nvidia Feynman GPU.
• Whether Intel Diamond Rapids and Clearwater Forest-related Foveros Direct 3D capacity expansion resumes in 2027.
• Whether Samsung HBM4E adopts 16-hi, and whether the initial number of hybrid bonding layers is only one.
• SK hynix and Micron's plans to introduce hybrid bonding for HBM5 in 2028.
• Besi's actual share of logic and HBM hybrid bonding equipment, and the pace at which Korean domestic equipment vendors enter the market.
• DISCO's order trends in advanced packaging, NAND CBA, BSPDN, and the recovery of SiC investment.