🧠 Food for thought

Implications, context, and why it matters.

The Dell deal is the public face of a broader AI storage strategy

• SK hynix bases its 321-layer QLC NAND push on the AIN D (Density) product line, which aims to pack more data into the same physical footprint 1.
• The company aims for petabyte-scale SSDs that could take over from hard disk drives when storing huge AI datasets 1.
• After this PC drive, SK hynix plans to bring 321-layer NAND to enterprise SSDs for data centers and to Universal Flash Storage (UFS), an embedded storage format widely used in smartphones 2.
• Compared with the prior generation, the technology delivers up to 56% faster writes and over 23% better write power efficiency, which fits the power limits of AI data centers 2.

SK hynix runs two NAND development tracks inside the company

• SK hynix is rolling out 321-layer QLC NAND for its own SSD roadmap, while Solidigm, its US-based subsidiary acquired from Intel, has shipped 192-layer QLC 3D NAND in its highest-capacity enterprise drives 3.
• Solidigm is working on future products such as a planned 245TB SSD using floating-gate NAND, which stores data by trapping electrical charge in a conductive layer; it also lists charge-trap NAND, which traps charge in an insulating layer, for other products in its portfolio 3.
• Together, the two groups can explore more than one NAND design while targeting the same end market, replacing nearline hard drives, high-capacity disks used for data that does not need the fastest access speeds, in AI infrastructure 1.
• Dell says tight storage integration matters for enterprise AI rollouts, though its blog post does not call for “extreme density” or frame the shift as a direct challenge to hard disk drives in bulk storage 4.