Blog

DEEP RESEARCH · SAMSUNG ELECTRONICS MEMORY CYCLE

Reading Samsung Electronics Through the Memory Super Cycle

A chart-based note comparing the 2016-2017 cycle with the possible 2025 AI inference cycle

Date: 2025-10-08 · Chart/cycle perspective · Naver Blog

Investment decisions are your responsibility. This material is research and is not a recommendation to buy or sell.

0. Bottom line first

If Samsung Electronics is trying to break above a roughly four-and-a-half-year range, I do not see it as just a chart event. The demand cycle behind the breakout matters first.

Interpretation: The 2016-2017 move was tied to mobile and data-center demand, 2020 to temporary COVID-era demand, and 2025 may be tied to storage demand from the AI inference cycle. The 2025 cycle still needs time to be confirmed.

Repeated Breakout PatternDemand shock → supply constraint → price/share response
2016-2017Phone demand and Galaxy Note 7 replacement demand
From 30,000 won range to 58,000 won
2020Temporary COVID demand
From 60,000 won range to 95,000 won
2025AI inference and storage demand hypothesis
After 95,000 won: unknown
WatchpointNew highs need an industry and supply-demand backdrop
The key is not the chart alone, but the memory demand and supply constraint behind it.

1. Three Demand Cycles Visible in the Chart

Official fact: The source notes that Samsung Electronics is attempting to break above its prior high and that the latest range appears to have lasted about four and a half years.

Long-term Samsung Electronics chart with range breakout areas
PeriodDemand backdrop in the sourcePrice areaRead
2016-2017Rising phone demand, with replacement demand after the Galaxy Note 7 overheating issueBreakout from 30,000 won range to 58,000 wonA cycle where mobile/data-center demand met supply constraints
2020Likely temporary demand increase from COVIDBreakout from 60,000 won range to 95,000 wonPandemic demand reflected in memory pricing and share price
2025Potential storage demand from the AI inference cycleAfter breaking 95,000 won: unknownA period to test whether AI infrastructure lifts memory and storage demand

2. Comparing 2016-2017 With the 2025 AI Cycle

Official fact: The source describes the current market as similar to the mid-2017 “peak supply constraint” phase. It cites that most 2025-2026 HBM capacity has already been sold and that 2026 general-purpose DRAM undersupply could be 23%, more than twice the 11% HBM shortage estimate.

Interpretation: AI data-center demand may be more structural than prior server replacement cycles, but simultaneous capex by the three major suppliers raises the risk of oversupply and price correction in 2026-2027.

2016

Demand Shift

Demand moved from PC DRAM toward mobile and data-center NAND, and DRAM contract prices turned upward after 18 months.

2024-2025

HBM Bottleneck

AI accelerator demand made HBM the bottleneck, while DRAM capacity conversion toward HBM tightened legacy supply.

2026+

Expansion Risk

HBM4 competition, Samsung P4/P5 investment, and NAND storage demand may unfold together.

3. Memory Demand Map After 2026

Official fact: The source highlights NVIDIA Rubin, Rubin GPUs R100/R200, Rubin CPX, 128GB GDDR7, SOCAMM2, and high-capacity NAND as demand streams after 2026.

Memory Demand Splits Across AIThis is no longer only an HBM story
HBM4Rubin GPU R100/R200, generation phase
GDDR7Rubin CPX, context phase, 128GB
SOCAMM2AI PCs and modular servers
High-capacity NANDAI data storage demand
Growth potential must be tracked together with hyperscaler capex slowdown and HBM oversupply risk.

4. Related Reading

Related cycle note: https://blog.naver.com/star_of_self/224034851149

Preview image for related Samsung cycle post

Sources