Samsung BM1743 QLC Enterprise SSD 61.44TB

The Samsung BM1743 is a sixth-generation QLC (quad-level cell) enterprise SSD engineered to address the exponential capacity demands of modern AI infrastructure, data lakes, and read-intensive scale-out storage deployments. Built on Samsung's latest V-NAND QLC technology with an advanced in-house controller, the BM1743 achieves a remarkable 61.44TB of raw capacity in a 2.5-inch U.3 form factor, enabling data centers to consolidate petabyte-scale datasets into dramatically fewer drive bays and reducing both rack footprint and power consumption per terabyte compared to traditional HDD-based storage tiers.

The drive's architecture is specifically optimized for sequential read throughput, sustaining up to 14GB/s reads over a PCIe Gen 5 x4 interface, making it ideally suited for AI training data ingestion, large language model checkpoint storage, and analytics pipelines that demand rapid, parallel access to massive datasets. Samsung's dynamic SLC caching and advanced wear-leveling algorithms ensure that write operations remain predictable under sustained workloads, while the embedded power-loss protection circuitry and end-to-end data path protection safeguard data integrity across all operating conditions.

As a purpose-built read-intensive enterprise drive, the BM1743 carries a workload rating optimized for read/write ratios exceeding 90:10, reflecting the operational profile of data lake tiers, object storage backends, and video surveillance archives. Its enterprise-grade reliability features—including TRIM, T.10 DIF/DIX end-to-end data protection, sanitize instant erase, and full AES 256-bit hardware encryption—position it as a secure, compliant solution for regulated industries. The combination of QLC density economics with enterprise firmware maturity makes the BM1743 a landmark product for hyperscale operators and enterprise IT organizations seeking to maximize usable capacity per rack unit.

Ideal for

• AI and machine learning data lake storage: providing high-density, high-throughput repositories for training datasets consumed by GPU clusters at scale
• Large language model (LLM) checkpoint archiving: storing frequent model snapshots with fast sequential read retrieval to minimize GPU idle time during training restarts
• Hyperscale object storage backends: replacing HDD-based capacity tiers with all-flash density to improve latency and reduce the number of nodes required per petabyte
• Video surveillance and media archive: enabling dense, always-available storage for high-resolution video streams with sustained read bandwidth for simultaneous playback
• Analytics and data warehousing cold-warm tiers: accelerating query performance on infrequently written, frequently scanned datasets in columnar analytics environments
• Content delivery and streaming origin storage: consolidating large media libraries into minimal physical footprint while sustaining the read throughput required for concurrent streaming sessions

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.