Electronics Business Strategy Briefing

December 10, 2025

TOPPAN Holdings Inc.

  1. Overview of Electronics Business

  2. Overview and Strategy of Semiconductor Packaging Business

  3. Trends in Focus Areas of Semiconductor Packaging Business

  4. Technology Roadmap for Semiconductor Packaging Business

  1. Overview of Electronics Business

  2. Overview and Strategy of Semiconductor Packaging Business

  3. Trends in Focus Areas of Semiconductor Packaging Business

  4. Technology Roadmap for Semiconductor Packaging Business

Business Overview

Electronics

FY2024 results



TOPPAN Holdings

Consolidated net sales

Living & Industry

FY2024 results

16%

of total sales

Net sales: JPY 283.3 billion Non-GAAP operating profit: 53.4 billion (Operating margin: 18.9%)

Semiconductors

  • Semiconductor packaging

  • Etched products

  • LSI design services

Sales

JPY 194.8 bn

Non-GAAP

operating

margin

Approx. 27%

  • Anti-reflective films

  • Light control films

  • TFT-LCDs

  • Color filters

Displays

Sales

JPY 88.5 bn

Non-GAAP

operating margin

Approx. 2

Information & Communication

FC-BGAs are expected to further recover in H2 due to an increase in the proportion of products for AI and the

contribution of the new line at the Niigata Plant

35.8

40.0

150

35.5

30.0

100

20.0

50

10.0

0

0.0

FY22 FY23 FY24 FY25 FY25

(Revised) (Original)

Displays

Operating Profit

Semiconductors

Non-GAAP OP

200

50.0

49.6

48.2

50.0

53.0

250

50.3

49.8

53.4

(¥bn)

60.0

283.3 190.0 259.0

Net sales

255.3 266.5

(¥bn)

300

Electronics

- Net sales by sub-segment & Operating profit(RHS) -

Category

Full Year Sales Composition

H2 Forecast

Full Year Operating Profit Margin

(YoY change)

GAAP

Non-GAAP

Approx. 77%

(-48.9 billion yen)

  • Photomasks: No sales or operating profit will be posted following the transition of Tekscend Photomask to the equity method

Semiconductors

  • Continuing the trend for Q2, the proportion of FC-BGAs

for server CPUs and network switches is expected to increase. We expect profit levels to rise significantly in Q4 due to tapping into strong demand with the full-scale operation of the new line. Qualifications for high-end AI server switches and AI ASICs are progressing in line with expectations in preparation for mass production in the next fiscal year.

Approx. 24%

(-16.8 billion yen)

Approx. 24%

(-16.8 billion yen)

Displays

Approx. 23%

(-44.4 billion yen)

  • Anti-reflective films: H2 is expected to see a recovery trend due to taking in demand for high-value-added products

  • Display solutions: Profit is expected to increase due to the effects of structural reforms

Approx. 3%

(-0.2 billion yen)

Approx 3%

(-0.3 billion yen)

Special factors

  • Foreign exchange: Profit decrease of approx. 2.8 billion yen over the full year

  • Impact of bonus provision period standardization (-0.6 billion yen)



Electronics: FY2025 Forecast



Focus area: Semiconductor packaging business

Focus on high-end FC-BGA market centered on AI applications

Launch advanced semiconductor packaging business

Anti-reflective films: Launch new ultra-wide line in 2026

Quantum dots: Launch nanomaterial

business centered on materials

Light control films: Expand business of new products for automotive applications

Execute business-specific strategies and portfolio transformation

Semiconductor Business

Display Business

Drive structural reform of low-profit businesses

LSI design: Generate synergies with the semiconductor packaging business

Etching: Expand into new fields such as

application to heat exchangers

The Electronics business will accelerate expansion by concentrating management resources on the high-value-added business of semiconductor packaging and driving portfolio transformation

Firmly maintain high profitability by supplying cutting-edge key devices leveraging technical superiority centered on semiconductor-related technologies

Sales

CAGR

Sales Approx. JPY 350 bn

Guidance

Excl.

Tekscend Photomask

22%

Sales

CAGR

Semiconductor-related portion

Percentage of semiconductor

-related sales: Approx. 80%

Sales

JPY 190 bn

Semiconductor-related portion

Sales

JPY 129 bn

Sales

JPY 146 bn

Non-GAAP

operating margin

24%

Semiconductor-related portion

Growth Drivers

  • Semiconductor packaging business

    26% Sales

    Approx.

    JPY 270 bn

    Non-GAAP

    operating margin

    30%

    Sales Growth Potential

    Profitability

    Semiconductors

    CAGR

    26% Non-GAAP

    Operating

    Margin 30%



    FY2025

    FY2025

    Enhanced profitability accompanying expansion of AI-related applications for FC-BGAs

    Sales JPY 85 bn

    Non-GAAP

    operating margin

    26%

    Launch and scaling of advanced semiconductor packaging

    FY2030

    INDEX
    1. Overview of Electronics Business

    2. Overview and Strategy of Semiconductor Packaging Business

    3. Trends in Focus Areas of Semiconductor Packaging Business

    4. Technology Roadmap for Semiconductor Packaging Business

    5. Specific Activities in Semiconductor Packaging Business

    © TOPPAN Holdings Inc. 8

    About FC-BGAs

    • FC-BGA (Flip Chip-Ball Grid Array) substrates are high-density semiconductor packaging substrates that enable high-speed, multifunctional LSI chips

      Product Cross-section

      IC chip



      FC-BGA

      Business Opportunities



      • Network devices, server CPUs Generative AI Consumer, automotive devices

By developing substrates with ultra-high-density interconnect structures through our evolution of microfabrication and build-up wiring board technologies, we provide products that support semiconductor

Breakdown of FC-BGA market and TOPPAN's business by application

2024 FC-BGA Market (value basis)

TOPPAN's FC-BGA Business in 2024 (value basis)

Consumer, etc.

25%

High-end switches,

servers, AI applications

83%

Consumer, etc.

Source: TOPPAN estimates based on data from Fuji Chimera Research Institute

We are advancing a business centered on high-end switches and servers

TOPPAN's FC-BGA Business Domains



FC-BGA Substrates for High-end Switches: Fiscal 2024 Sales by Manufacturer



A Co.

B Co.



C Co.

D Co.

E Co.

F Co.

G Co.

H Co.

FC-BGAs for high-end switches

Share: No. 3

Source: TOPPAN estimates based on data from Fuji Chimera Research Institute

TOPPAN has secured the no. 3 market share in the high-end switch domain

TOPPAN's Position in the High-end Switch Domain



Focus areas targeted by TOPPAN

: Supported: Focus area -Not supported

High-end

switches

Servers

General-purpose

For AI

AI accelerators

Server CPUs

GPUs

AI ASICs

X86

(Intel)

ARM

-

-

Focus areas:

  1. High-end switches

  2. AI ASICs

  3. Server CPUs

    AI

    accelerators

    High-end switches Server CPUs

    PC CPUs

    Gaming

    Automotive, etc.

    AI ASICs

    GPUs

    Focus areas

    With technical superiority as the source of our competitive edge, we are targeting high-end areas where we can leverage our advantage

    Request

    Response

    General-purpose switches

    ・・ ・

    ・・ ・

    ・・ ・

    Network centered on

    general-purpose servers

    Switch for AI

    Server CPUs

    AI ASICs

    Network centered on AI servers

    AI switch

AI server

AI server

AI server

AI server

General-purpose server

AI server

General-purpose server

Switch

Switch

Switch

Switch

Switch



Specific composition of focus areas

High-end switches

Servers

General-purpose

For AI

AI accelerators

Server CPUs

GPUs

AI ASICs

X86

(Intel)

ARM

-

-

Focus areas are ARM server CPUs, AI ASICs, and switches for AI, which are expected

to be used in AI data centers

Market forecast for focus areas

(Billion JPY) FC-BGA Substrate Market

High-end switches

AI accelerators Server CPUs

TOPPAN's focus areas

18,000

1,800

15,000

1,500

12,218

1,221.8

17,343

1,606.7

14,199

16,067

1,419.9

18,256

1,734.3

1,825.6

Server CPUs

High-end switches General-purpose/AI

AI accelerators ASICs GPUs

12,000

1,200

9,687

10,304

968.7 1,030.4

x86

Server CPUs

ARM

9,000

900

6,000

600

3,000

300

Level of difficulty

Automotive, consumer, etc.

PC CPUs

PC GPUs

Networks (NIC, etc.)

Gaming

Automotive

0

2023 2024 2025 2026 2027 2028 2029

(Source: TOPPAN estimates based on data from Fuji Chimera Research Institute)

Networks

Middle/low-end

Projected panel unit price

(Source: TOPPAN estimates based on data from Fuji Chimera Research Institute)

High-profit/high-end focus areas are expected to expand against a background of growth in AI

Proposing solutions geared towards social issues

Creating products and solutions encompassing all facets of semiconductor packaging

Interposer

Advanced FC-BGA

Photonics-electronics

convergence support

Resin / Glass core

Identifying new needs and advancing development by bringing together our technologies and collaborating with customers, partner companies, and universities

  • Collaboration with research institutions/universities

  • Identification of technology needs: US-JOINT, alliance with Taiwanese companies

  • North American fabless customers

  • Our technologies:

    FC-BGA micro interconnect/layering technology,

    glass transfer, LSI design, process clean-up, CMP slurry (TOPPAN Infomedia)

  • Development bases:

Semiconductor Packaging Development Center (Sugito), Ishikawa Plant

  • Existing Niigata FC-BGA line: Online

  • New Niigata FC-BGA-line: Full-scale startup in January 2026

  • AST Phase 1 line: Mass production launch scheduled for fiscal 2027

  • Advanced semiconductor packaging at Ishikawa: Startup scheduled for July 2026

    Mass production scheduled for fiscal 2028 or later

  • New Ishikawa FC-BGA line:

    Mass production scheduled for fiscal 2030 or later

  • AST Phase 2 line: Timing TBD

*Plan is current projection

Expanding business scale*

INDEX
  1. Overview of Electronics Business

  2. Overview and Strategy of Semiconductor Packaging Business

  3. Trends in Focus Areas of Semiconductor Packaging Business

  4. Technology Roadmap for Semiconductor Packaging Business

  5. Specific Activities in Semiconductor Packaging Business

© TOPPAN Holdings Inc. 16

AI Semiconductor Market Expansion and Focus Areas

(Billion USD)

Proportion of Global Semiconductor Demand Accounted for by AI Semiconductors



The AI semiconductor market is projected to grow

1,200

by 7x (CAGR 28%) over the 8 years up to 2030 and account for 70% of the total market in 2030

1,005

CAGR

28%

1,000



844

920

800

600

564 526

618 673

717

776

AI Applications Driving the Market

400

200

AI smartphones

Driving

Cloud



automation

Humanoids

Edge



0

2022 2023 2024 2025 2026 2027 2028 2029 2030

(Source: IBS)

Servers

© TOPPAN Holdings Inc.

AI is driving semiconductor market growth, and cloud demand

centered on servers is also increasing 17

  • The emergence of AI has kickstarted an era in which all kinds of things generate data (data-centric), and data traffic is projected to increase 100x in the 20 years from 2020 to 2040

    Change in Data Traffic Towards 2024

    (Zettabyte)

    35

    30

    25

    20

    15

    Data traffic projected to increase 100x in the 20 years to 2040



    Data-

    CAGR

    26%

    10

    5

    1990

    Computer-centric

    2000

    Mobile-

    centric

    2010

    centric

    2020

    2030

    2040

    (Source: OMDIA)

    Accelerated Increase in Data Traffic Against Backdrop of AI



  • AI processing involves constant synchronization while processing across multiple GPUs, meaning the volume of communication within and between servers via switches is greater than on networks centered on conventional general-purpose servers



  • This is driving a transition to network architectures that enable scaling-up, to enhance the performance of individual AI servers, and scaling-out, to facilitate distributed processing with multiple servers connected in parallel

    Network centered on general-purpose servers

    Method for enhancing processing performance by raising server performance and expanding hardware functions, such as CPUs and memory

    Switch

Switch

Scale-up

Switch

Switch

Method for enhancing processing performance by increasing the server count and stabilizing operation via distributed processing

AI server

AI server

Network centered

on AI servers

Scale-out

・・ ・

・・ ・

・・ ・

AI switch

AI server

AI server

AI server

Switch

AI server



Request Response

Spine Leaf

General-

purpose server

General-

purpose server

General-

purpose server

General-

purpose server

General-

purpose server

General-

purpose server

Network switch count and performance requirements increase to process vast

data volumes

Changes in Network Architecture



Data center switch market size forecast

汎用向 け

平均単 価

250,000

200,000

150,000

AIサーバ 向け

出荷金 額

  • AI-related applications are driving market expansion due to increased demand stemming from server parallelization and unit price increases prompted by performance enhancements

    (Source: Fuji Chimera Research Institute's 2025 market survey on data centers, AI, and key devices)

    2030 (Year)

    2029

    2028

    Projection

    2027

    2026

    2024 2025

    Forecast

    2030 shipment volume for AI is

    6,000 projected to be 3.2x the 2024 level

    5,000

    4,000

    3,000

    2,000

    1,000

    0

    For general-purpose

    For AI servers

    7,000

    (1,000 units)

    Data Center Switch Shipment Volume

    (Source: Fuji Chimera Research Institute's 2025 market survey on data centers, AI, and key devices)

    0

    2030 (Year)

    2029

    2027 2028

    Projection

    2026

    2025

    Forecast

    2024

    50

    0

    5,000

    100

    10,000

    150

    15,000

    250

    200

    20,000

    300

    Shipment value Average unit price

    25,000

    Shipment value

    (Billion JPY)

    Shipment Value

    Data Center Switch Unit Selling Price/

    Average unit

    price (10k JPY)

    350



    100,000

    50,000

    Switches for AI servers will drive market expansion

    Network Switch Demand Trend



    AI(推論用) (1,000台)

    AI(学習用) (1,000台)

    Change of phase in terms of what is required of AI processing

    Computing

    power important

    Large

    number of chips connected

    Memory

    speed

    important

    A few chips

    are sufficient

    Use of pre-trained foundation models

Foundation model development

Inference

Learning



(Source: Fuji Chimera Research Institute's 2025 market survey on data centers, AI, and key devices)

2030 (Year)

2029

2028

Projection

Forecast

2027

2026

2025

2024

0

1,000

2,000

3,000

4,000

AI (learning (1,000 servers)

AI (inference) (1,000 servers)

AI Server Count Projection

  • Computation required for AI is classified into two types: Training and Inference

  • The accumulation of pre-trained foundation models will drive progress in the shift to inference-based applications

AI-related needs will shift from learning to inference

ALU

Semiconductor trend accompanying expansion of AI inference

Arithmetic

operations

Memory

Software-controlled staged computation

CPU/GPU

Input

Output

Power savings achieved by optimizing

and shortening interconnects

Acceleration with reduced software control

Custom Circuit (ASIC)

Output

Input



  • ASICs are circuits designed for specific applications-they are expected to deliver processing that

    is faster and more energy efficient than GPUs for general-purpose applications

  • From the perspective of speed and efficiency, there are increasing needs to select chips based on application-with GPUs being used for model learning and AI ASICs used for inference applications that use pre-trained foundation models

    Power consumption efficiency is required for inference processing

    Market trend for in-house AI ASIC production volume

    (Source: Fuji Chimera Research Institute's 2025 market survey on data centers, AI, and key devices)

    2030 (Year)

    2029

    2028

    Projection

    2027

    2026

    2025

    Forecast

    2024

    0

    2,000

    4,000

    6,000

    8,000

    CAGR

    16%

    Projected In-house AI ASIC Production Volume

    (1,000 units)

    10,000



    © TOPPAN Holdings Inc.

    With the growth of AI inference, demand will increase for AI ASICs

    2031

    with superior power consumption efficiency 23

    0

    (Source: Fuji Chimera Research Institute's 2025 market survey on data centers, AI, and key devices)

    0%

    2031 (Year)

    2030

    2029

    2028

    Projection

    2027

    2026

    2025

    Forecast

    2024

    (1,000 units)

    50,000

    10,000

    10%

    20,000

    30,000

    20%

    40,000

    ARM x86 Percentage of ARM

    30%

    Projected Server CPU Shipment Volume by Type



    ARM系

    x86系

    ARM系比 率

    • There are x86 CPUs, used by companies such as Intel, AMD, and IBM, and ARM CPUs, used by companies such as NVIDIA and cloud vendors like GAFAM

    • x86 CPUs previously had an overwhelming share of the market, but in recent years the share of ARM CPUs produced in-house by cloud vendors has been increasing, particular for AI servers with high power consumption

      © TOPPAN Holdings Inc. 24

      INDEX
      1. Overview of Electronics Business

      2. Overview and Strategy of Semiconductor Packaging Business

      3. Trends in Focus Areas of Semiconductor Packaging Business

      4. Technology Roadmap for Semiconductor Packaging Business

      5. Specific Activities in Semiconductor Packaging Business

25

© TOPPAN Holdings Inc.



Advanced semiconductor packaging

Chiplet

(Silicon interposer + FC-BGA)

Glass core FC-BGA

Organic RDL interposer + FC-BGA

Glass interposer + FC-BGA

Ishikawa Plant, CapEx

Start of full-fledged mass production Ishikawa Plant

Establishment of Development Center

Development Center

FC-BGAs continue to evolve along with chiplets, driving structural diversification

FC-BGA substrate

Support for IOWN2.0

IOWN2.0

Board-level optical

interconnection

FC-BGA (board-level optical interconnection)

FC-BGA (chip-to-chip optical interconnection)

Niigata Plant, new line

FC-BGA

FC-BGA



(large-body, high-count layer)

IOWN

concept

Niigata Plant, new line

Singapore, plant construction and CapEx

Larger chips

The IOWN concept promoted by NTT, Inc. aims to utilize optical technologies to achieve

a low-power, high-quality, high-capacity and low-latency network.

IOWN3.0

Chip-to-chip optical

interconnection

IOWN4.0

On-chip optical interconnection

Start of mass production

Singapore

Start of mass production

Niigata Plant

Mass production begins on extended line

Niigata Plant, existing line

From 2030 onwards

2029

2028

2027

2026

2025

2024

2023

2022

2020

CY

Sustainable Growth Phase

Result Delivery Phase

Foundation Building Phase

Management Phase

"IOWN®" is a trademark or registered trademark of NTT, Inc.

With our new technologies, we contribute to meeting increasingly sophisticated demand for

TOPPAN's Technology Roadmap



Projected Transmission Speed Required of AI Switches (TOPPAN estimate)

(Tbps)

Use Cases Requiring High Speed and Low Latency

900

800

700

600

500

400

300

200

100

0



409.6

819.2

204.8

102.4

25.6 51.2

AI ASICs

Server CPUs

Request Response





Driving automation Telehealth

AI server

AI server

AI server

2020 2022 2024 2026 2028 2030 (year)

Network centered on AI servers

Switch

Switch

Switch

Switch

Switch

AI server

AI server

AI switch

Switch for AI

・・ ・

・・ ・

・・ ・

General-purpose

high-end

switches

AI-centric next-generation use cases, such as driving automation and telehealth, require high-speed, low-latency transmission-the transmission speed of data centers and the like is forecast to reach 800 Tbps in 2030, 32x the 2020 level

Next-generation AI Network Transmission Performance Projection



Miniaturization of interconnects is vital to achieve chiplet structures that integrate heterogeneous chips

  • To address the narrowing pitch of I/Os, Moore's Law is complemented by maximizing interconnect

    density and ensuring high bandwidth

  • Power efficiency is increased by minimizing parasitic capacitance and reducing die-to-die RC characteristics (resistance, capacitance)

  • Reducing the system's physical size allows for miniaturization and a smaller form factor.

Unit: Micron

Features

Application

2018

2019

2020

2021

2022

2025

2028

2031

2034

Min. line width/space (um)

FC-BGA

9/12

9/12

9/12

8/8

8/8

5/5

5/5

5/5

5/5

Fan-out, Organic

interposer

2/2

2/2

2/2

1.5/1.5

1.5/1.5

1/1

1/1

0.5/0.5

0.5/0.5

Source: Heterogeneous Integration Roadmap (2021 Edition), Chapter 8 Section 8 Table 1. Substrate interconnect scale roadmap

Interconnect miniaturization is expected to enhance value in terms of Performance, Power, and Area, with interconnect dimensions below L/S=1µm/1µm required in 2030

Pursuit of Miniaturization



Conventional

Single chip semiconductor

Single SoC

80 mm

Semiconductor chiplet for AI applications

>200 mm

I/O chips Chiplet SoC

HBM (high bandwidth memory) Optical element

Large interposer

Large-body FC-BGA

Package substrates are becoming larger to enable the large-scale integration of heterogeneous chips.

  • With conventional AI/switch semiconductors, performance enhancements have been achieved through single-chip miniaturizing and scaling up, but the industry is now facing challenges related to fabrication complexity and cost increases

  • As a breakthrough to meet demand for further performance enhancement, chiplet technology-which involves dividing

and reconfiguring semiconductor chips into individual "chiplets"-will become the mainstream technology going forward

Co-packaging of large-scale integration and components such as optical elements-

impossible with single chips-is supported by the scaling of packaging components

Semiconductor Packaging Structure Trend

Future (2030)



Conventional

Future

Silicon interposer

Next-generation interposer

活用

ケー ス

従来

シリコンインター ポーザー

Use cases

HBM and SoC high-density connection

HBMとSoCの高密度接続

⇒伝送帯域確保のためにイ ンターポーザを活用

~ Use interposer to ensure transmission bandwidth

SoC + HBM + IO+ SerDes chip + optical element, etc.

SoC+HBM+IO+SerDesチップ+光素子 等

⇒大規模な異種チップ混載 実現のためにインターポーザを活用

将来

次世代インター ポーザー

~ Use interposer to achieve large-scale integration of

heterogeneous chips

大型 化

Scaling-up

微細 化

Miniaturization

Wafer processing adopted due to use of silicon materials

シリコン材料のためウエ ハプロセス採用

⇒将来の大型対応は事実上 不可

~ Support for future scaling is effectively impossible

ウエハプロセス採用のため微細化に優れる。

L/S=0.5µm

Suited to miniaturization due to wafer processing L/S = 0.5 μm

Base material with high die count such as large square panel required Organic materials are compatible with the concept of manufacturing in square form

大型角パネルような高取り 数の基材が求め

られる。⇒有機材料は角型生 産の思想に合致

シリコンインターポーザ ーに肉薄する

微細化性能を持つ必要が ある

Needs to have miniaturization performance close

to that of silicon interposers

For the massive scaling of chiplet packages, the industry is looking to next-generation interposer technologies that support larger sizes as an alternative to silicon interposers

Interposer Trend



Attachments

  • Original document
  • Permalink

Disclaimer

Toppan Holdings Inc. published this content on December 10, 2025, and is solely responsible for the information contained herein. Distributed via Public Technologies (PUBT), unedited and unaltered, on December 10, 2025 at 06:16 UTC.