ReportGem ReportGem 中文

REAL-TIME GLOBAL RESEARCH

Advantest (6857.T): Beyond the Cycle: Strong tester demand led by GPUs/ASICs, but valuations also still look high

Published: 2026-06-15Institution: Goldman SachsPages: 7Original language: EnglishEvidence page: 2

Research evidence excerpt

Advantest (6857.T): Beyond the Cycle: Strong tester demand led by GPUs/ASICs, but valuations also still look high

Goldman Sachs Advantest (6857.T)

Drivers of SoC tester demand

Testers for GPUs account for the largest proportion of the CY26 SoC tester TAM

(Advantest assumes US$8.7-9.5 bn), but ASIC demand is also rising substantially as AI

semiconductor demand shifts from training to inference, and the future growth rate for

ASICs is expected to be higher than that for GPUs. Management’s view is that the TAM

for ASICs could exceed that for GPUs in three to four years. Demand for CPUs for

inference applications is also expanding considerably, and while the company stands to

benefit if they are Arm-based CPUs, management said the market size may not be as

large as that for GPUs/ASICs from the perspective of the number of cores and

transistors. Regarding SoC tester market share, the company noted that while there are

signs of competitors looking to enter the GPU space, the situation has not changed

significantly.

Future new growth drivers

Potential future growth drivers cited by Advantest included testers for CPO

(co-packaged optics), die-level probers that test singulated chips, system-level

testing (SLT) with the aim of further penetrating the AI/HPC space as it pivots from its

previous focus on smartphone applications, and test interface boards. Regarding

testers for CPO, while customers have yet to establish testing processes, Advantest’s

policy is to leverage the expertise it has gained through c.20-30 years of developing

optical measuring instruments to expand sales of solutions for high-density optical

devices as well.

Price Target Risks and Methodology - Advantest

Valuation methodology: Our 12-month target price of ¥30,000 is based on the global

The English excerpt is extracted automatically from the cited source page and may contain layout or recognition errors. It is never batch translated.

Open report viewer