SK Chairman Says Company May Offer ‘Memory as a Service’
THE SO WHAT
Turning DRAM into “memory as a service” is a move to pull part of the value chain up into recurring revenue and tighter integration with compute. If you run memory-bound AI or analytics workloads, start asking how pricing, latency, and data locality would change if your vendor exposes memory as an on-demand resource instead of a component.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechSK Chairman Chey Tae-won on SK Hynix Debut, AI Demand and US Plans
A $26.5 billion US listing by a memory pure-play is a bet that AI-driven demand for HBM and DRAM is durable, not a blip. If you’re planning large model or inference footprints, assume memory pricing and availability are now strategic variables, not afterthoughts.
Deep & Emerging TechThornburg's Di Zhou Sees Memory Supercycle
Calling a "memory supercycle" with SK Hynix as a core holding underscores how central HBM and DRAM have become to the AI trade. Infra buyers should expect investors to reward capacity expansion and long-term offtake deals, which may lock in pricing power upstream.
Deep & Emerging TechSK Chairman Says He Has 'Much, Much Bigger' Plans for US
SK Group telegraphing larger US investments is another data point that advanced memory and battery supply chains are localizing around US incentives. Operators building AI or EV capacity should map where SK’s future fabs and plants land, because those sites will anchor talent and ecosystem gravity.
Deep & Emerging TechSK Chairman Has 'Some Confidence' Demand Will Grow
When SK Group’s chair says customers are actively asking them to expand, that’s a signal that hyperscaler and AI infra demand curves still look steep from inside the supply chain. If you’re planning large GPU or HBM-heavy projects, assume continued tightness and lock in capacity early.