Yes, the NAND data from TrendForce is extremely bullish, but the key question is whether this is a short squeeze cycle (2025-2026) or a structural supercycle (to 2027+).
---
1. Why NAND prices are exploding now
TrendForce recently raised its Q1 2026 NAND price forecast to +85–90% QoQ, driven by a severe supply-demand imbalance.
Key drivers:
AI infrastructure demand
Hyperscalers are buying massive amounts of enterprise SSDs for AI training and inference.
AI workloads need large vector databases and extremely high-IOPS storage.
Capacity shift
Memory manufacturers are prioritising enterprise SSD and server products, reducing supply for consumer markets.
Supply discipline
NAND demand is projected to grow 20–22% annually while supply rises only ~15–17%, widening the shortage.
Result: suppliers currently hold strong pricing power.
---
2. Why Micron earnings could surprise
Memory stocks are the most operationally leveraged in semiconductors.
When prices rise:
Revenue increases immediately.
Gross margins expand dramatically.
With NAND + DRAM both surging:
Analysts expect memory prices to stay tight through 2027 due to AI demand.
Even new fabs (Micron Taiwan expansion etc.) will not add meaningful capacity until late decade.
This is why memory stocks often outperform late in semiconductor cycles.
---
3. Will the memory “supercycle” extend to 2027?
Possibly, but it will look different from past cycles.
Structural demand drivers
1. AI data centres
Massive SSD storage for AI training datasets.
2. AI PCs
Copilot PCs require more local storage and memory.
3. AI smartphones
Larger models need higher NAND capacity and faster storage.
4. Edge AI devices
Cameras, robots, autonomous systems.
TrendForce expects memory revenue growth to continue through 2027 due to these trends.
---
4. But there are still cycle risks
Memory is historically the most cyclical semiconductor sector.
Possible brakes:
Consumer device slowdown (smartphones already weak).
Oversupply once fabs ramp in 2027–2028.
Hyperscaler capex slowdown.
In other words:
2026: explosive pricing
2027: tight but stabilising
2028+: potential oversupply cycle
---
5. Market implication (for investors)
Memory may be one of the most underowned AI trades.
Semiconductor AI stack:
Layer Leader
AI compute Nvidia
Networking Broadcom
Memory Micron, SK Hynix, Samsung
Storage SanDisk, Kioxia
If AI compute keeps expanding, memory is mathematically required. Every GPU cluster consumes enormous DRAM and NAND.
---
✅ My view:
The memory cycle is real, not just hype.
The strongest period for memory stocks is typically mid-cycle when prices spike.
The market may still be underestimating NAND demand from AI inference storage.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
