Blog

DEEP RESEARCH · LB Semicon

LB Semicon: from a DDI house to a SoC test hub

What Exynos 2600, HPB, and the LB Lusem merger imply for the Korean OSAT supply chain

Published: 2025-12-26 · Semiconductor back-end research · Original Naver Blog post

You are responsible for your own investment decisions. This material is research, not a recommendation to buy or sell.

0. Bottom line first

The core view in the source is that the Exynos 2600 wafer-test win can re-rate LB Semicon from a DDI-heavy OSAT into a higher-value SoC test company.

Official fact: The source says LB Semicon was selected in November 2025 as a Samsung Exynos 2600 wafer-test partner. It also treats the February 1, 2025 absorption merger with LB Lusem as a key financial and process event.

Interpretation: Moving beyond a business where DDI represented 70-80% of revenue can change customer mix, technical difficulty, and valuation multiples at the same time.

Legacy

DDI 70-80%

The traditional revenue base was tied to the display cycle.

New pillar

Exynos 2600

A symbol of entry into high-reliability AP wafer testing.

Merger

2025.02.01

The LB Lusem merger strengthens turnkey bumping, test, and packaging capability.

1. Front-end change, back-end opportunity

LB Semicon source image 1

The report frames Samsung Foundry's 2nm SF2 process and HPB as comeback cards after 3nm yield issues and the Exynos 2500 mass-production failure. Exynos 2600 production and the 2026 Galaxy S26 scenario are the starting point for a shift in Korean system-semiconductor back-end work.

Samsung non-memory back-end flowWhere LB Semicon newly participates
Fabless/IDMSystem LSI design
FoundrySF2 production
OSATBumping, wafer test, packaging
End customerGalaxy S26 and others
Higher process difficulty can lift test time and ASP.

Official fact: The source says Samsung Foundry began stabilizing 2nm SF2 yields at the 50-60% level from the second half of 2025 and entered Exynos 2600 production.

Interpretation: 2nm and HPB are not only front-end stories. More test items and longer test time can improve OSAT unit pricing and utilization.

2. Business model and merger synergy

LB Semicon receives finished wafers, forms bumps that connect chips to substrates, tests electrical performance, and sorts known-good die. Gold bumping is the traditional cash cow for LCD/OLED DDI, while solder and Cu pillar bumps are used for AP, PMIC, CIS, and other high-performance non-memory chips.

ProcessRoleMeaning in the source
BumpingForms Au, solder, or Cu protrusions on the dieElectrical connection to substrate or leadframe
Wafer testUses probe cards to test electrical characteristicsThe Exynos 2600 CP test win falls here
Back-end assySawing, grinding, WLCSP and related workCombined with LB Lusem's packaging capability

Official fact: The merged company is described as a KRW 769.1bn-asset OSAT as of Q3 2025. LB Lusem had strengths in display COF packaging and power-semiconductor back-end processing.

Interpretation: Combining LB Semicon's bumping/test with LB Lusem's packaging/assembly enables turnkey offerings. The merged company's liquidity also supports Exynos test equipment and Anseong expansion CAPEX.

3. Customers and 2023-2025 investment flow

Key customers include LX Semicon, Samsung Electronics, MagnaChip, and Himax. LX Semicon is the major fabless company in the LG Display ecosystem and a large DDI bumping/test source, while Samsung System LSI is the growth axis for Exynos and ISOCELL.

2023

CIS line

Anseong added a CIS test line and cleanroom expansion.

2024

AP testers

Even during the DDI downturn, the company began ordering high-end AP test equipment.

2025

Dedicated setup

After the Exynos 2600 win, Exynos test equipment setup and readiness followed at Anseong.

4. Galaxy S26 scenarios and utilization

The source assumes annual Galaxy S26 series shipments of about 35 million units and about 15-20% Exynos wafer-test share for LB Semicon as a new vendor.

ScenarioExynos adoptionUnitsLB Semicon utilization read
Bear20%About 7 millionLimited effect if yield remains unstable
Base30-40%About 10-14 millionSoC test line could recover to about 75% in H1 2026
Bull50%+About 17.5 million85%+ utilization and possible expansion pressure

Official fact: Current Q3 2025 non-memory line utilization is described as the mid-60% range because of weak DDI demand.

5. Nepes comparison, HPB, and risks

Nepes began in 1990 as a semiconductor materials company and became a first-generation OSAT. In the Exynos ecosystem it has been tied to PMIC, WLP, and FO-PLP. Nepes Lawe, spun off in 2020, invested more than KRW 600bn in a 600mm panel-based FO-PLP facility but fell into capital impairment at end-2023 after yield and customer delays, according to the source.

HPB reduces the heat trap from DRAM stacked above the AP by using a copper heat block or placing AP and DRAM side by side through FOWLP. The source says thermal resistance can fall by more than 30%, while early main AP packaging is likely to be led by Samsung AVP, leaving OSAT opportunities in related components, sub-packaging, and PMIC.

LB Semicon source image 2
Valuation or riskSource numberCheckpoint
LB Semicon PBR0.7-0.8x on 2025 expected earningsWhether weak DDI is already priced in
EV/EBITDAAbout 7.7x versus global OSAT average of about 8-10xMultiple expansion from SoC test entry
Re-rating range2026 PER below 10x, possible PBR 1.2-1.5xExynos volume and merger-effect visibility
Risks2nm yield, heat/performance issues, delayed IT set recoveryGalaxy S26 Exynos share and DDI utilization

Interpretation: The source classifies LB Semicon as Top Pick, Nepes Ark as Interest, and Nepes as Watch. I read that as a distinction in risk-adjusted exposure, not as a standalone investment instruction.

Sources