IBM researchers find a way to store 3 bits in a phase-change memory cell, paving the way for memory technology with DRAM-like speed and a cost closer to flash
Flash storage is too slow for your device's main memory, but RAM is expensive and volatile. Thanks to a breakthrough from IBM …
Context & Ripple Effects
IBM's 3-bits-per-cell phase-change result lands mid-race: over the previous year, Intel and Micron had already put 3D Xpoint into production claiming 1,000x flash speed without transistors, while HP and SanDisk bet on their own next-generation storage chip partnership predicting a similar 1000X jump. The shared target is the gap between fast-but-expensive volatile DRAM and cheap-but-slow flash.
IBM's contribution is density: packing 3 bits into one phase-change cell borrows the multilevel-cell trick Samsung used to push 3-bit NAND into smartphones, but applies it to a material that switches fast enough to sit near main memory. If it scales, one technology could serve both tiers instead of two.
First-order effects
- IBM now has a credible lab-stage answer to Intel-Micron's shipping 3D Xpoint and HP-SanDisk's announced program, turning storage-class memory from a two-horse race into three competing material approaches.
- Samsung and other flash makers face a technique that could let a rival match flash-like cost per bit while escaping flash's speed ceiling.
Second-order effects
- Server and device designers gain a candidate for collapsing the separate DRAM-plus-storage bill of materials into one memory tier, pressuring pricing across both markets if any approach reaches volume.
- The multilevel-cells-per-bit arms race shifts from NAND alone to every emerging memory type, since bit density is what makes non-DRAM technologies cost-competitive.
Third-order effects
- If phase-change, Xpoint-style, or memristor approaches all mature, the industry's hard split between 'memory' and 'storage' vendors and product lines starts dissolving into a single storage-class memory category.
- Whoever controls a universal memory wins leverage over system architecture itself — the same consolidation pressure Samsung is exercising from the DRAM side with high-capacity server modules would come from below on cost.
The trend: Vendors are converging on storage-class memory that erases the DRAM-versus-flash tradeoff, with IBM's phase-change work joining Intel-Micron and HP-SanDisk as competing bets on which material gets there first.