ESDS Launches High-Performance GPU-as-a-Service Platform
ESDS Software Solution Limited has introduced a high-performance GPU-as-a-Service platform. Announced during the company’s 20th Annual Day, the service targets the increasing compute needs of machine learning, large language models, and generative workloads for enterprises, financial institutions, research organizations, and government bodies.
With this launch, ESDS positions itself as a full-stack provider, offering cloud, managed services, data center infrastructure, and software solutions, now complemented by large-scale, sovereign-grade GPU systems.
The company stated that the new service is designed to provide high-performance computing at a global scale, catering to organizations with demanding computational requirements.
This announcement comes as global investment in high-performance servers, including GPUs and accelerators, is projected to reach $329.5 billion by 2026, fueled by the need for predictable, high-throughput computing platforms.
ESDS’s platform allows organizations to execute critical workloads on custom-built GPU SuperPODs engineered for secure operation, consistent performance, and low-latency distributed training.
Over time, the company has developed its expertise into a fully managed GPU infrastructure stack that enables organizations to scale complex workloads on a reliable and well-architected foundation.
Piyush Somani, promoter, managing director, and chairman of ESDS, explained that this initiative addresses the growing demand for large-scale compute infrastructure. He said the service makes access to extensive GPU clusters and SuperPODs simple, transparent, and tailored for enterprises pursuing advanced computing.
Somani added that ESDS’s GPU SuperPODs “offer predictable performance, stability, and scalability,” and highlighted the SuperPOD Configurator tool, which allows organizations to select GPU models, design their clusters, and instantly visualize architecture and costs.
Advanced GPU Systems and Enterprise Services
The offering includes high-performance GPU systems such as NVIDIA DGX, HGX B200, B300, GB200, NVL72, and AMD’s MI300X platforms. These systems are optimized for training very large models, accelerating computations, running simulations, and handling extensive data operations.
ESDS said the GPU SuperPODs leverage high-bandwidth NVLink, unified memory pools, intelligent scheduling, improved thermal management, and orchestration tuned for workloads to deliver consistent performance at scale.
The service portfolio includes consulting for dedicated GPU clusters, deployment and supply of GPU environments, GPU infrastructure-as-a-service, hybrid CPU+GPU cloud options, and fully managed on-demand GPU services.
ESDS takes responsibility for architecture design, network optimization, container orchestration, performance tuning, and round-the-clock monitoring with full operational support.
The SuperPOD Configurator enables enterprises to design infrastructure by choosing GPU models, compute density, memory configurations, storage options, and interconnect setups, automatically producing optimized architecture, performance projections, and cost estimates.
The company shared an example of a research lab that reduced training time for a 50-billion-parameter model from over 40 days to 10 days, cut costs by 60%, and achieved 30× faster processing by switching to NVL72-based systems with optimized containers and high-speed interconnects. ESDS emphasized that the platform meets global performance standards but is designed and optimized in India, serving over 1,300 enterprise, financial, and government clients with transparent pricing, flexible consumption models, and integrated services.
Also Read:
DeepSeek Joins OpenAI and Google in Achieving Gold at IMO 2025
CloudExtel Raises ₹200 Crore to Expand Data Centre Interconnect Network
