PCIe Gen5 SSDs Move From Showcase to Standard
Across the show floor, PCIe Gen5 SSDs emerged as the new baseline for performance‑class systems, especially AI workstations and creator PCs. KIOXIA’s XG10 client NVMe SSD illustrates how quickly the standard is maturing: moving to a PCIe 5.0 x4 interface and NVMe 2.0d, it delivers up to 14,000 MB/s sequential read and 12,000 MB/s write, plus up to 2,000K / 1,600K random read/write IOPS. Compared with its PCIe 4.0 predecessor, the XG8, KIOXIA claims roughly 2x faster sequential throughput and well over 100% gains in random performance, positioning the drive for AI‑assisted applications, content creation, and high‑end gaming. addlink’s virtual showcase echoes this trend with the G57 PCIe Gen5 SSD, also rated up to 14,000 MB/s, and the more mainstream G55 targeting demanding everyday workflows. Together, these launches signal that PCIe Gen5 SSDs are no longer niche components but core infrastructure for next‑generation client and workstation platforms.
DDR5 Memory Targets AI Servers, Workstations, and Intelligent PCs
Memory makers used the event to pivot decisively toward AI workloads. addlink’s SC5 DDR5 CUDIMM pushes speeds up to 8,400 MT/s, using an on‑module clock buffer to maintain signal integrity at high frequencies in upcoming platforms such as Intel Arrow Lake‑based systems. G.SKILL, long known for enthusiast overclocking kits, is now foregrounding server‑grade RDIMM, workstation RDIMM, and ECC UDIMM lines designed for AI servers, enterprise workstations, and advanced industrial PCs. The company has already been supplying overclockable RDIMMs to high‑speed computing environments at research institutions, and is now explicitly targeting corporate AI infrastructure and on‑premises data processing. V Color is following a similar trajectory, expanding from gaming into AI servers and AI workstation hardware with memory capacities scaling from 16GB to 256GB to support flagship accelerator platforms. DDR5 memory is thus evolving from a speed upgrade into a foundational element of AI workstation hardware and data‑centric compute nodes.
Thunderbolt 5 Storage and AI Accelerators Blur the Lines Between Compute and Capacity
High‑speed external connectivity is becoming an extension of system memory and storage, rather than just a peripheral bus. OWC showcased a full portfolio of Thunderbolt 5 storage and connectivity solutions, including the Envoy Ultra, a certified 8TB Thunderbolt 5 portable SSD, and StudioStack, which offers up to 32TB of stackable Thunderbolt 5 storage for creative and production workflows. The headline, however, is OWC Stack AI: billed as the world’s first Thunderbolt 5 AI accelerator and storage hub, it expands effective AI working memory for existing Windows and Linux systems, enabling larger models to run locally without hitting GPU VRAM limits. By combining acceleration and storage in a single Thunderbolt 5 enclosure, OWC points toward a future where Thunderbolt 5 storage functions as both fast media and an AI compute companion, reducing dependence on cloud inference and giving organizations greater control over performance, privacy, and cost.
Hybrid Storage Hubs and NVMe Docking Stations Redefine Portable Workflows
Vendors are increasingly merging high‑speed storage with docking and hub capabilities to support mobile creators and edge AI users. addlink’s B31 Magnetic NVMe Docking Station exemplifies this shift, combining an integrated NVMe SSD enclosure with PD 100W charging, HDMI 4K60 output, USB‑C, USB‑A, and SD/TF card slots in a single compact device. For laptop‑first workflows, it offers a true one‑cable setup that handles both expansion storage and I/O, especially when paired with addlink’s P30 USB4 portable drive, capable of up to 4,000 MB/s. Verbatim is exploring a similar hybrid direction with new Pocket SSD models that integrate hub functionality—adding HDMI, USB‑A, and USB‑C ports—plus magnetic attachment for mobile devices and, on some units, write‑protection to guard against viruses and accidental deletion. Together, these NVMe docking station and hub‑SSD designs show how storage is becoming the backbone of minimalist, high‑bandwidth desks and field rigs.
From Storage‑Power Integration to Cloud‑to‑Edge AI Ecosystems
Beyond raw bandwidth, Computex highlighted how storage and memory are being woven into broader AI and power ecosystems. Verbatim’s S’n’P (Storage and Power) concept ties advanced external SSDs to semi‑solid power banks and GaN chargers, underscoring how reliable data access now depends on equally robust energy delivery. ADATA, through its TRUSTA enterprise brand and industrial division, presented a full cloud‑to‑edge AI hardware roadmap. Its AI Scaler memory‑storage solution and toolkit dynamically orchestrate GPU, DRAM, and SSD resources to reduce dependence on high‑end GPUs, with claimed deployment cost reductions of over 50% for AI training and inference. At the edge, ADATA demonstrated PCIe Gen5 SSDs, AI PCs, and NVIDIA Jetson Thor‑based robotics using digital twin integrations for smart healthcare. Paired with G.SKILL’s and V Color’s AI server and workstation memory portfolios, the show made clear that storage‑power integration and distributed, edge‑ready AI infrastructure are fast becoming central industry trends.
