Innovative Island Cloud

Innovative Island Cloud (IIC) is the latest large-scale artificial intelligence and cloud computing platform constructed by the National Center for High-Performance Computing (NCHC), launched in August 2026. Integrating high-performance computing (HPC) with cloud technology, the entire IIC system aims to accelerate the multimodal development and practical application of Taiwan's generative AI foundation models. Powered by the next-generation NVIDIA H200 NVL GPU accelerators, the platform delivers powerful high-speed computing, scalability, and data transfer capabilities. It is paired with 15PB of storage space, offering diverse services including Block Storage, Object Storage (e.g., S3), and File Sharing. IIC provides highly flexible on-demand self-service resources suitable for industry, government, academia, and research sectors.

IIC features an integrated, unified user portal, allowing researchers to effortlessly switch between different regions within a single interface to manage resources. It also supports the offline cross-region migration and scheduling of virtual machines. To safeguard system and computing security, in addition to providing security groups and isolated virtual network protections, the platform integrates an underlying cybersecurity monitoring platform equipped with semantic understanding and automated threat analysis. This AI-collaborative defense ensures the security of sensitive data and computing environments.

As a versatile and highly flexible AI cloud platform, IIC offers custom-configured virtual machines, Auto Scaling, and Load Balancing features, empowering users to build small-scale computing clusters and run distributed AI model training across multiple nodes. IIC also provides Kubernetes cluster services (K8s) and out-of-the-box application services such as Jupyter Notebook, fully meeting the diverse needs of AI training, big data analytics, and cross-disciplinary scientific research applications.

Innovative Island Cloud (IIC) is a highly flexible AI computing platform supporting a multi-region architecture, capable of hosting over 19,200 cloud virtual machines simultaneously. Currently, two regions are available for selection: the AI-Cloud Region located in Tainan, and the AI-Trust Region located in Taichung. The server hardware specifications for each region are as follows (see the detailed table below):

AI-Cloud System Specification

CPU nodes	· 2x Intel Xeon 6767P 2.4GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 6400 RDIMM*32
GPU nodes	· 2x Intel Xeon 6767P 2.4GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 6400 RDIMM*32 · NVIDIA H200 NVL 141GB PCIE*2

AI-Trust System Specification

CPU nodes	· 2x Intel Xeon Platinum 8592V 2.0GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 5600 RDIMM*32
	· 2x Intel Xeon 6767P 2.4GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 6400 RDIMM*32
GPU nodes	· 2x Intel Xeon Platinum 8592V 2.0GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 5600 RDIMM*32 · NVIDIA H100 80GB PCIe*2
	· 2x Intel Xeon 6767P 2.4GHz CPU (64 Cores/CPU) · Main memory 64 GB DDR5 6400 RDIMM*32 · NVIDIA H200 NVL 141GB PCIE*2