Today for AI
HOT RADAR
computeHEAT 10.9°

AI CLUSTERED EVENT · 10/12/2026

Samsung Develops zNAND-T and zNAND-I Flash: Optimizing Throughput and Random IOPS for AI Workloads

1 reports archived1 independent sourcesupdated 10/12/2026, 13:54:37
Synthesis & Latest Updates
1 Sources Cross-Validated

Samsung is developing two new NAND flash products, zNAND-T and zNAND-I, specifically optimized for AI workloads to address bandwidth bottlenecks in large model inference. zNAND-T utilizes TSV stacking to enhance data center read/write throughput, while zNAND-I focuses on optimizing random performance for frequent small-data operations, particularly suitable for KV Cache offloading. This shift indicates that NAND flash is evolving from general-purpose products to high-performance custom solutions tailored for specific AI tasks.

LATEST/Samsung is developing two new NAND flash products, zNAND-T and zNAND-I, specifically optimized for AI workloads to address bandwidth bottlenecks in large model inference. zNAND-T utilizes TSV stacking to enhance data center read/write throughput, while zNAND-I focuses on optimizing random performance for frequent small-data operations, particularly suitable for KV Cache offloading. This shift indicates that NAND flash is evolving from general-purpose products to high-performance custom solutions tailored for specific AI tasks.

HEAT TRENDHourly heat curve

4 fully observed hours
Now
10.9
Peak
11.715:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

TIMELINECoverage timeline

Total 1 reports · Latest first
  1. IT之家 · 智能时代T2·68 pts
    • Samsung introduces zNAND-T and zNAND-I, two new flash memories optimized for AI workload throughput and random performance respectively.
    • zNAND-T uses TSV-stacked NAND Dies, similar to the HBF concept, aiming to boost read/write bandwidth for data center AI semiconductors.
    • zNAND-I emphasizes performance in frequent small-data reads/writes, primarily targeting KV Cache offloading scenarios, reflecting a trend toward customized storage.
Samsung Develops zNAND-T and zNAND-I Flash: Optimizing Throughput and Random IOPS for AI Workloads | Today for AI