NETWEB announced an expansion to its range of NVIDIA MGX platform based servers. India's first ‘Make in India' Tyrone servers, incorporating NVIDIA MGX-based Grace and NVIDIA GH200 GraceHopper Superchips, will be offered by Netweb for off-prem and on-prem AI cloud infrastructure. This will be combined with allied software and services for generative AI and scientific innovations. Netweb currently designs and manufactures more than ten models under its flagship Tyrone range of NVIDIA MGX AI systems.
Netweb's exemplary work on end-to-end design, PCBA, production and building block architecture expertise has advanced the R&D, manufacturing and solutions capabilities for AI in India. Netweb's long-standing relationship with NVIDIA now includes building India's own sovereign AI compute infrastructure with both off-prem and on-prem models.
The new range of Tyrone AI systems accelerated by NVIDIA Grace and Grace Hopper Superchips is powered by a wide range of NVIDIA and Netweb's respective software stacks to build the AI cloud infrastructure for off-prem and on-prem scenarios.
Tyrone Skylus cloud instances can be created and managed by users and organizations using and taking advantage of the broad software stack, which includes NVIDIA AI Enterprise, NVIDIA Omniverse, Skylus.ai, KubytsHub and KubytsFlow. This stack introduces cloud instance autonomy leveraging hundreds of pre-compiled/trained frameworks – models, and development tools to structure the development, training and deployment process over AI cloud. This will play a substantial and essential role in Generative AI , large-scale agentic AI and physical AI workloads, helping users to scale across hundreds and thousands of nodes in an AI cloud.
Netweb's NVIDIA MGX systems with different configurations of GPUs, CPUs, and DPUs—including x86 or Arm® CPU servers and NVIDIA OVX™ servers—are offered focusing on building block and rack-scale architecture. They also offer both direct- to-chip liquid-cooled and air-cooled designs. With direct-to-chip liquid cooling technology, Tyrone AI systems will help build a large AI cloud with much higher TDP limits for GPUs and CPUs, yet maintaining the best PUE for the AI cloud infrastructure.
Netweb's comprehensive range of Tyrone NVIDIA MGX, x86 architecture-based GPU systems for end-to-end AI cloud solutions for CSPs and enterprises, will be demonstrated at the NVIDIA AI Summit at the Jio Convention Centre in Mumbai, 24 and 25 October 2024.