GIGABYTE (2376-TW) subsidiary GIGASTEEL today announced the expansion of the GIGAPOD computing and infrastructure product combination, launching multiple modular POD design solutions to accelerate the deployment speed of global enterprises in AI factories, AI training and inference platforms, and high-performance computing (HPC). GIGASTEEL mentioned that this infrastructure is built on GIGABYTE's server technology, integrating leading accelerated computing, high-speed networking, storage, and software ecosystem, aiming to reduce the deployment complexity and operational risk for enterprises when landing large-scale AI applications. The newly expanded GIGAPOD product combination can support large language model (LLM) training, retrieval-augmented generation (RAG), autonomous agent AI, large-scale inference, digital twins, and sovereign AI projects, among other diverse workloads. To meet the needs of enterprises continuously expanding the scale of AI applications, this platform provides pre-configured architecture design and OEM integrated deployment services, effectively shortening the time required to produce tokens and significantly reducing the time and resources required for basic infrastructure to go online. In terms of hardware configuration and cluster management, this solution adopts a scalable architecture that can correspond to small to large-scale multi-rack AI clusters and provides air-cooling and liquid-cooling basic infrastructure options, as well as high-performance network options supporting InfiniBand and Ethernet switches. At the same time, the system integrates GIGABYTE's G-REX and GPM dual software management ecosystems, centrally managing power, cooling, and real-time resource monitoring through a 1U management node, and can immediately notify administrators through the interface when liquid leakage is detected, further establishing a high-reliability and controllable operation foundation.
FACT BOX
- Source: PR Times
- Category: New Product