Latest Search
Quote
| Back Zoom + Zoom - | ||||
|
<Research> Nomura Raises Samsung Electronics (005930.KS) TP to KRW0.59M, SK Hynix (000660.KS) TP to KRW4M; AI-Driven Memory Demand Shows Exponential Growth
Recommend 17 Positive 39 Negative 25 |
||||
|
||||
|
Nomura said in a report that AI-driven demand is growing exponentially while memory supply remains limited, and expects memory stocks to undergo valuation re-rating. The broker raised the TP for Samsung Electronics (005930.KS) to KRW590,000 from KRW340,000, and lifted the TP for SK Hynix (000660.KS) to KRW4,000,000 from KRW2,340,000. Both are rated Buy. The report noted that the memory industry entered a structural growth phase after the launch of ChatGPT in December 2022, triggering significant growth in demand for GPUs and HBM. Subsequently, adoption of retrieval-augmented generation (RAG) and agentic AI applications accelerated demand for traditional servers and SSDs, and since 3Q25 has kicked off a triple memory supercycle (commodity DRAM, HBM and SSD). Nomura said that as AI semiconductor demand shifts from training to inference workloads, memory demand is entering a period of exponential expansion. This is reflected in rapid growth in KV-cache memory demand, driven by factors including user base, user engagement time, AI task complexity, increased inference token consumption, and the emergence of agentic AI (with token usage per user surging exponentially). Memory demand is effectively a multiplicative function of these variables (10x200x30x10x&), which the broker believes implies that memory demand could grow by several thousand times over the next five years. In contrast, it expects industry supply growth over the same period to be constrained to around 5x to 6x (with a CAGR of about 30%), raising serious questions over whether the structural supply-demand imbalance can truly be resolved. The broker added that the industry is currently attempting to narrow the widening supply-demand gap through various software and architectural optimizations. However, Nomura believes these solutions can only slow the pace of growth rather than reverse the trend. As such, it expects every viable approach including NAND offloading (which, although slower, can increase capacity by more than 100 times) and adoption of ultra-high-bandwidth NAND may need to be deployed simultaneously. Meanwhile, the rise of agentic AI is significantly enhancing AI productivity while also driving a substantial increase in token consumption. Agentic AI is also creating entirely new categories of semiconductor demand, including CPUs and commodity memory. However, Nomura believes these developments will ultimately reinforce the expansion of AI server usage itself, thereby driving another round of substantial growth in GPU and memory-related demand. (da/) Auto-translated by AI This article was automatically translated by AI, the original language version should be considered the authoritative version. AASTOCKS.com Limited does not guarantee its accuracy or completeness and accepts no liability for any damages or losses arising from the use of this translation. More Details
AASTOCKS Financial News |
||||
Copied to Clipboard
 |
||||
