SK Hynix PCB01 PCIe Gen5 Enterprise SSD 2TB

The SK Hynix PCB01 is a PCIe Gen 5 x4 NVMe 2.0 enterprise SSD engineered for AI-PC and edge-server deployments demanding extreme sequential throughput at low latency. Built on SK Hynix's 238-layer 4D NAND flash technology and paired with an in-house controller, the PCB01 delivers sequential read performance reaching up to 14,700 MB/s and sequential write performance up to 13,000 MB/s in the 2 TB capacity point, positioning it among the fastest client-to-edge class drives available at its launch in late 2024. The drive operates in the M.2 2280 form factor, making it suitable for space-constrained AI inference nodes, edge servers, and high-performance workstation builds where PCIe Gen 5 slots are available.

The PCB01 carries the manufacturer part number SHGP5-2000GM and is designed to address the storage bottlenecks that arise in AI inferencing pipelines, large model loading, and real-time data preprocessing at the edge. Its architecture prioritises bandwidth headroom so that CPU and GPU resources are not starved waiting on storage I/O, which is a common constraint in on-device AI workflows. The drive supports NVMe 2.0 command set features including host memory buffer (HMB) optimisation and advanced error correction to maintain data integrity under sustained workloads.

From a thermal and power perspective, the PCB01 incorporates dynamic thermal throttling management to sustain performance within the thermal envelope of compact edge and AI-PC chassis. End-point security is supported through hardware-level AES 256-bit encryption, ensuring data-at-rest protection compliance for enterprise and regulated-industry deployments. The combination of Gen 5 bandwidth, 238-layer 4D NAND density, and the M.2 form factor makes the PCB01 a practical upgrade path for organisations modernising edge infrastructure ahead of expanding on-device AI workload requirements.

Ideal for

• On-device AI model loading and inference acceleration in edge servers and AI-PC nodes where rapid access to multi-gigabyte model weights is critical to inference latency
• High-throughput data preprocessing pipelines that stage raw sensor, video, or telemetry datasets before GPU-accelerated processing at the edge
• Enterprise workstation storage for large-scale CAD, simulation, and generative AI creative workloads that saturate PCIe Gen 4 storage bandwidth
• Edge content delivery and media processing nodes requiring sustained sequential read throughput for 4K/8K video transcoding and streaming tasks
• Compact edge server OS and application fast-boot volumes where rapid cold-start times reduce downtime and improve availability in distributed deployments
• Secure data-at-rest environments in regulated industries such as healthcare or financial services where AES 256-bit hardware encryption is a compliance requirement

Technical specifications

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