SK Hynix PEB110 PCIe Gen5 Enterprise SSD

The SK Hynix PEB110 is a flagship enterprise NVMe SSD built on the PCIe Gen5 x4 interface, representing SK Hynix's latest generation of data center storage engineered specifically for the throughput and latency requirements of AI infrastructure, large-scale model training, and hyperscale cloud deployments. Leveraging SK Hynix's advanced 238-layer 4D NAND flash technology paired with a purpose-designed Gen5 NVMe controller, the PEB110 achieves sequential read speeds exceeding 14 GB/s and sequential write performance surpassing 10 GB/s — nearly doubling the throughput ceiling of previous PCIe Gen4 enterprise drives. The drive's architecture is optimized for sustained, consistent performance under heavy queue depths typical of AI training pipelines, checkpoint I/O, and high-concurrency inference serving environments.

The PEB110 is offered in the industry-standard U.2 (2.5-inch) and E3.S form factors, making it fully compatible with current and next-generation server platforms from major brands supporting PCIe Gen5 NVMe. The drive incorporates enterprise reliability features including end-to-end data path protection, power-loss protection (PLP) capacitors, and advanced thermal management with dynamic throttling algorithms that preserve rated performance within standard 25W thermal envelopes. Wear leveling and error correction leverage LDPC ECC technology tuned for high-density 4D NAND, delivering consistent endurance ratings suitable for write-intensive workloads.

Designed for deployment in AI servers, GPU compute clusters, and disaggregated storage architectures, the PEB110 addresses the acute I/O bottleneck that emerges when large language models and multimodal AI systems must stream training datasets and model checkpoints at scale. Its combination of raw bandwidth, microsecond-class latency, and enterprise-grade reliability positions it as a foundational storage component for organizations building or expanding AI-optimized data center infrastructure in 2025 and beyond.

Ideal for

• AI and large language model training clusters requiring sustained high-bandwidth sequential data streaming to GPU memory
• Model checkpoint storage in distributed deep learning environments where rapid write throughput reduces training cycle overhead
• High-throughput inference serving platforms demanding low-latency model loading and rapid context retrieval
• Hyperscale cloud object storage tiers requiring dense, high-bandwidth NVMe drives in disaggregated storage nodes
• Real-time analytics and vector database workloads processing large embedding indices with mixed sequential and random I/O
• High-performance computing scratch storage for simulation and rendering pipelines that saturate PCIe Gen4 drive bandwidth

Technical specifications

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