ELSA Japan has announced the GigaIO Gryf, a portable edge AI system built into a carry-on suitcase that brings high-performance computing directly to on-site environments without relying on traditional data center setups.
The system focuses on flexibility and mobility, which allows teams to deploy AI workloads in remote or space-constrained locations while keeping the total weight under 25 kg depending on configuration.
Modular design with powerful hardware
The Gryf system uses GigaIO’s FabreX PCIe and CXL fabric, which lets users reassign CPU, GPU, and storage resources through software without changing cables, making it easier to adapt the system based on workload needs. It includes six internal sled slots where users can mix compute, accelerator, storage, and networking modules.
The platform supports AMD EPYC processors and NVIDIA GPUs like L40S and H100 NVL, with storage scaling up to 246TB NVMe, while SWARM linking connects multiple units for higher performance and larger storage capacity.
