AI giants "must buy"; manufacturers hold the greatest pricing power! Morgan Stanley: This round of the memory "supercycle" will far exceed historical peaks.

Morgan Stanley believes that a new AI-driven "super cycle" for memory has already arrived, with its intensity and logic fundamentally different from any previous cycles.

According to Wind Trading Desk, Morgan Stanley stated in its latest report that DRAM prices are breaking historic highs, ushering in an unprecedented "super cycle."

Morgan Stanley points out that, unlike before, this cycle is led by AI data centers and cloud service providers, whose customers are less sensitive to price, with inference workloads now the primary driver of general memory demand. On the other hand, recent channel research shows that Q4 server DRAM contract prices have surged nearly 70%, NAND contract prices are up 20-30%, and suppliers have gained unprecedented pricing power.

The bank maintains its outperform rating for SK Hynix and Samsung Electronics, expecting rising memory prices to drive stock prices to new highs, and memory makers’ profits to significantly exceed expectations.

AI Demand Changes the Game — Clients "Have No Choice but to Buy"

Morgan Stanley notes the core driving force of this cycle has fundamentally changed. The demand side is no longer price-sensitive traditional clients, but AI data centers and cloud giants waging an arms race to build computing infrastructure.

For them, acquiring memory is a strategic "necessity," with price sensitivity minimized. Meanwhile, HBM (High Bandwidth Memory) production is structurally squeezing out traditional DRAM capacity. As emphasized by Morgan Stanley in its report:

The difference lies in: today’s memory demand has evolved into a competition dominated by AI data centers (compute-intensive platforms) and cloud service providers, who are much less price-sensitive than traditional customers... The exponential growth trend of inference demand provides a solid foundation, making this cycle fundamentally different from any before.

Unprecedented Price Surge, Suppliers Hold Absolute Power

On the other hand, Morgan Stanley says that with DRAM contract prices soaring, suppliers now have unprecedented pricing power.

Their latest channel research shows that DRAM price prospects have strengthened sharply in just two weeks. Q4 server RDIMM contract prices have surged nearly 70%, far exceeding the previous 30% forecast. DDR5 (16Gb) spot prices rocketed 336%, rising from $7.50 in September to the current $20.90. DDR4 prices have also jumped 50%. Although most contract deals will be completed later this month, customer acceptance seems inevitable—they fear further price increases and supply limitations.

Looking further, NAND is facing a severe shortage:

NAND has become a key component for AI compute infrastructure and video storage. 3D NAND wafer (TLC and QLC) prices are expected to rise 65-70% quarter-over-quarter due to constrained capacity. Nearline storage specs are switching from 128TB to 256TB QLC SSDs. TrendForce projects that enterprise SSD bit demand for servers will grow nearly 50% year-on-year in 2026. In the first half of 2025, Samsung’s bit output will be constrained by the transition from V6 176-layer to 321-layer V8-NAND, with only gradual ramp-up in the second half, resulting in just a 10% increase in bit shipment volume this year.

Room for Prices to Rise Further, Cycle Far From Peaking

The market is often constrained by "fear of heights," believing that new highs in stock prices mean a reversal is near. But Morgan Stanley stresses that in this AI-driven rally, profit growth is the decisive factor, not historic valuations:

Currently, server DRAM is priced at $1/Gb, while the previous cloud super cycle in Q1 2018 peaked at $1.25/Gb. Considering the scale of AI infrastructure investment and the dynamic nature of hyperscale customers, this cycle’s peak pricing could surpass the last high. Memory cycles typically last 4-6 quarters, profits are being realized, but the key is the comparison with market expectations—the market appetite for general memory pricing is clearly much higher. Valuation isn’t a predictor of future returns; it reflects supply and demand, not historical precedent.

Currently, driven by robust long-term AI demand, rises in memory pricing have entered "uncharted territory," with profit outlook far exceeding market consensus, meaning there is still huge upside for stock prices.

“With ongoing expansion in AI-related capex, memory’s share in total spending may continue rising—this will bolster P/B far beyond historical peaks. In our view, the story is one of cyclical earnings recovery with valuation multiple expansion layered over it…

We believe analysts always adjust data too slowly—for Hynix and Samsung, our earnings forecasts for 2026 and 2027 are 31-48% and 38-51% higher than market consensus, respectively.”

Overall, the driving force behind this memory "super cycle" is more enduring, price increases surpass historical records, the profit outlook is clearly higher, and together with strong cyclical momentum, this creates a rare investment opportunity for memory makers with pricing power.

Risk Warning and DisclaimerThe market has risks, and investment needs to be cautious. This article does not constitute personal investment advice and does not take into account individual users’ specific investment objectives, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their particular circumstances. If you invest based on this, you are responsible for your own decisions.