IONOS has pioneered in the European cloud landscape by deploying the NVIDIA DGX H200 system within its ISO-certified data centres. This integration, facilitated through Cronon, marks a groundbreaking stride for companies, research institutions, and governmental bodies looking to harness advanced AI capabilities. The NVIDIA DGX supercomputing offers robust resources for developing multimodal AI applications, large language models (LLMs), and data-intensive simulations with scalability and energy efficiency.
With the deployment of NVIDIA DGX H200, IONOS seamlessly integrates the complete AI customer journey. From the AI Model Hub to dedicated GPU servers, businesses can now train and deploy AI initiatives adhering strictly to European legal and regulatory standards.
The DGX H200 transcends a single server component. It's a robust system with eight NVIDIA H200 Tensor Core GPUs housing 1128 GB of HBM3e memory, designed for extensive workloads. The GPUs are strategically connected via NVIDIA NVLink, offering a high-speed direct connection of 900 GB/s, complemented by NVIDIA NVSwitches delivering 7.2 TB/s bandwidth.
NVIDIA ConnectX-7 NICs provide impressive 400 GB/s network bandwidth. Completing this powerhouse setup are dual Intel Xeon Platinum 8480C processors, totaling 112 cores, 2 TB of system memory, and a 30 TB NVMe SSD. Such specifications cater to the AI community's need for powerful yet cost-effective training and inference solutions.
IONOS operates the Nvidia DGX H200 systems within its European data centres, ensuring compliance with stringent German and European data protection laws. This makes the platform exceptionally suitable for industries such as healthcare, finance, and public administration.
The impact of IONOS's deployment is palpable. European AI company Noxtua is using the DGX H200 to enhance legal AI for lawyers. The system is instrumental in training a data protection-compliant language model, positioning itself as one of the first sovereign legal AIs in Europe.