DEEP RESEARCH · SEMICONDUCTORS/EQUIPMENT & MATERIALS
Is a Legacy CAPEX Cycle Really Coming?
A review of the memory investment cycle and equipment/materials basket strategy based on an SK Securities IT equipment/materials report
0. Bottom line first
I still do not know whether a legacy semiconductor CAPEX cycle will really come. The SK Securities report, however, argues that 2025 marks another four-year investment cycle as AI training and inference drive high-performance semiconductor demand, with DRAM +54% and NAND +14%. I am also asking whether it is better to spend time studying this cycle or focus on continuously growing companies and irreplaceable equipment names.
1. My first reaction
I read the report today and, because I do not know the legacy cycle well, I am recording it as the start of studying it. I am not actually sure whether a legacy cycle will come, and part of me thinks simply holding Park Systems or Samsung Electronics may be enough.
Interpretation: I also wonder whether the time would be better spent studying industries and companies that keep growing without a cycle. In this cycle in particular, many factories are being built in the U.S., so I question whether they may use U.S. equipment vendors because maintenance is easier. If U.S. government pressure is also possible, investing around companies with irreplaceable equipment may be better.
2. Report core: a four-year investment cycle
Official fact: The source records an SK Securities IT equipment/materials report by Lee Dong-ju titled “IT Equipment/Materials (Overweight), Run, materials and equipment!” The report says memory investment is surging in 2025, following 2013, 2017, and 2021, because AI training and inference are driving explosive high-performance semiconductor demand. The stated numbers are DRAM +54% and NAND +14%.
3. Manufacturer investment roadmaps
Official fact: The source says memory manufacturers’ investment roadmaps for this year and next year are becoming more concrete. Samsung Electronics P4 DRAM is 60K/M, split into D1b 15K and D1c 45K. P4 NAND is V9 15K, with V10 orders in 2Q26. SK hynix M15X is scheduled for November completion at 60-70K/M.
| Company/line | Source number | Meaning |
|---|---|---|
| Samsung Electronics P4 DRAM | 60K/M, D1b 15K, D1c 45K | Concrete DRAM investment |
| Samsung Electronics P4 NAND | V9 15K, V10 orders in 2Q26 | Equipment order schedule to watch |
| SK hynix M15X | November completion, 60-70K/M | Checkpoint for the hynix investment cycle |
4. U.S. non-memory expansion and possible return to Korea
Official fact: U.S. non-memory expansion plans are also growing. The source notes an additional $100 billion for TSMC’s Arizona fab and more than $40 billion cumulatively for Samsung Electronics’ Taylor fab, with an initial 20K/M. It also mentions the possibility that the U.S. withdrawal of Samsung and SK hynix VEU status could lead investment back to Korea.
Arizona
An additional $100 billion investment plan is mentioned.
Taylor
More than $40 billion cumulative plan and initial 20K/M are presented.
Possible return to Korea
The U.S. withdrawal of Samsung and hynix VEU status may connect to a return of domestic investment.
5. How to view materials and equipment in a CAPEX cycle
Official fact: The report says basket buying of materials and equipment is the standard approach in a CAPEX cycle. Historically, equipment stocks formed bottoms at the early stage of Samsung Electronics memory CAPEX expansion, while materials and parts rose with a 3-6 month lag. The conclusion is that sentiment is improving across the materials/equipment chain.
6. Stock views and attached reports
Official fact: The source says Wonik IPS and Komico remain existing top picks, while Park Systems and Unisem are newly added top picks. It also records EO Technics and TES as recommended names with new momentum in materials and equipment.
| Stock | Source view/target price | Link |
|---|---|---|
| Wonik IPS | Buy, target price raised to KRW 55,000 | https://buly.kr/9tBKrgl |
| TES | Buy, target price raised to KRW 33,000 | https://buly.kr/2UjVfUl |
| Unisem | Buy, target price raised to KRW 11,000 | https://buly.kr/FhOLEgp |
| Park Systems | Buy, coverage initiated with KRW 330,000 target price | https://buly.kr/1xzEj3k |
| EO Technics | Buy, coverage initiated with KRW 270,000 target price | https://buly.kr/6XmwEHD |
The industry report URL is https://buly.kr/5q7uIul.
7. My checklist
- Whether the legacy CAPEX cycle actually turns into orders and earnings
- Whether U.S. fabs prefer U.S. equipment over Korean, Japanese, or European equipment
- Whether companies with irreplaceable equipment are a better choice than cyclical materials/equipment basket names
- Whether Park Systems, Samsung Electronics, and the report-mentioned companies are each worth studying
The conclusion remains open. The report’s numbers and cycle logic are worth studying, but I still need to compare whether buying the cycle or buying irreplaceable growth companies is the better choice.
Sources
- Original Naver Blog post: https://m.blog.naver.com/PostView.naver?blogId=star_of_self&logNo=223995748065
- Industry report: https://buly.kr/5q7uIul
- Wonik IPS company report: https://buly.kr/9tBKrgl
- TES company report: https://buly.kr/2UjVfUl
- Unisem company report: https://buly.kr/FhOLEgp
- Park Systems company report: https://buly.kr/1xzEj3k
- EO Technics company report: https://buly.kr/6XmwEHD