DEEP RESEARCH · Semiconductors / Memory Cycle

IM Securities Monthly Semiconductor (April) — Song Myung-sup's Bearish View and How I'm Responding

HBM density stuck, DRAM prices likely to roll over after Q2, my plan around hybrid bonding and glass substrates

Published: 2025-04-14 · Memory/equipment cycle analysis · Source: Naver Blog

0. Bottom line first

1. Summary of Song Myung-sup's bear case

Video link: https://youtu.be/tlWmnaMa2qo?si=TKMNuBJHGG1APvGy

Official fact (key points as relayed):

• Through Blackwell 200/300 and Rubin, HBM growth past 12-Hi is capped, so capacity expansion is limited. The next generation after Rubin "Cheap" is where density grows again → HBM capacity growth pessimistic through next year.
• Chinese smartphone inventory builds → DRAM prices likely roll over after Q2.
• Samsung's HBM plan called for +120% this year — risk is closer to 0% growth. If the plan misses, capacity reverts to legacy → legacy supply increases (Chinese players are also adding).
• Chinese demand uplift will be smaller than expected. Module players like Adata are stacking more inventory despite already being heavy — and there is some tariff-front-running.
• Q2 contract prices fix at end-April through customer negotiations. Micron's +10% and Samsung's +5% are memory makers' hopes; the agreed result is unknown.
• HBM equipment vendor numbers will likely deteriorate into Q2–Q3.
• If H2 was really a recovery setup, semis would already be turning — instead they have broken to new lows.
• Assume: cycle weak through H1 next year, recovery in H2 next year → then this year's H2 stock action could improve, but "we'll only know when we get there."
• Tariffs: temporarily 10%, but semis likely face component-level tariffs aimed at forcing US production — material drag.
• Global liquidity YoY has rolled and re-accelerated. China's subsidies — doubled this round — are losing punch as unemployment ticks up again.
• Samsung Electronics: KRW 55,000 = PBR 0.9 floor defensible. Quick rebound unlikely. In a recession/crisis scenario, PBR 0.75 = KRW 47,000 is open. Suitable for a one-year-horizon buy.
• SK Hynix is "more complicated."

2. How I'm responding

Interpretation / playbook:

• I can't go entirely empty at these prices. For semis exposure I am laying a base position via SOXX through retirement accounts; single-name calls wait until H2.
• The structural answer is HBM going higher stack count via hybrid bonding, plus glass substrates expanding substrate size → more HBM packages per device → real demand growth. That cycle is where momentum comes back.
• Targets to follow: hybrid-bonding leaders and companies solving the glass-substrate problems. I'm willing to be more flexible on valuation here — build positions gradually while learning.

3. Implications

• This cycle's next momentum will be next-gen packaging innovation more than raw HBM capacity additions.
• Macro variables (tariffs, Chinese demand, liquidity) drive near-term price action; technology lineup positioning matters separately for the next cycle.