Cartoon-like Semiconductor Rally: Samsung Electronics and SK hynix Big Catalysts Explode Simultaneously / The Identity of Nvidia's Six New Chips | Hyundai Motor Securities Research Center Director Noh Geun-chang

Summary

Noh Geun-chang, Director at Hyundai Motor Securities Research Center, discusses how Big Tech custom AI accelerators, Nvidia's six new rack-scale chips, and Microsoft's Maia 200 affect the semiconductor supply chain. He argues HBM, NAND/enterprise SSD, optical networking/silicon photonics, and Samsung Foundry are key beneficiaries, while glass substrate adoption may take longer. He also flags a likely first-quarter memory earnings surprise but says the rally may need a breather.

• Big Tech custom AI accelerators expand HBM demand for Samsung Electronics and SK hynix.
• Microsoft's Maia 200 uses 216GB of HBM, more than Amazon Trainium 3 and Google Ironwood.
• Nvidia's six new chips emphasize rack-scale design, CPO switches, DPUs, and NVLink networking.
• Silicon photonics and optical networking are seen as increasingly important for AI data centers.
• AI inference and KV-cache handling support enterprise SSD/NAND demand and tight supply.
• Samsung Foundry is improving on higher utilization, more customers, and advanced-node R&D.
• Glass substrate commercialization may be delayed because TSMC has not yet committed to it.
• First-quarter memory earnings could surprise on price hikes, though the stock rally may need rest.