Vultr has announced plans to deploy a 20 MW AI supercluster in Chennai, powered by AMD Instinct MI355X GPUs, to expand access to high-performance computing for AI development and deployment in India.
The deployment will mark Vultr’s largest AMD Instinct installation in India and give the cloud infrastructure provider two of the world’s largest AMD-powered AI infrastructure deployments, alongside its existing supercluster in Springfield, Ohio.
Each AMD Instinct MI355X GPU offers 288GB of HBM3E memory and memory bandwidth of 8TB/s, supporting compute-intensive workloads such as large language model (LLM) training, foundation model development, fine-tuning and high-throughput inference.
The Chennai facility is designed to serve AI research labs, foundation model developers, startups, enterprises, academic institutions and other organisations requiring large-scale AI computing resources.
J.J. Kardwell, CEO, Vultr, said the deployment would give Indian organisations access to local AI infrastructure to address data sovereignty, data residency and latency requirements while supporting the training, fine-tuning and deployment of AI models within the country.
AMD collaboration to expand AI computing
The Chennai project builds on Vultr’s collaboration with AMD across its Instinct GPU roadmap, including the MI300X, MI325X and MI355X GPUs, as well as planned support for the MI455X and AMD Helios rack-scale AI architecture. AMD EPYC processors also power Vultr’s VX1 Cloud Compute offering.
Vultr plans to expand its AI infrastructure capacity further in 2027, with additional deployments involving AMD Instinct MI455X GPUs and the AMD Helios architecture.
The Chennai supercluster will also use the AMD ROCm open software ecosystem, which supports widely used AI frameworks, libraries and models. The software platform is intended to give developers greater flexibility in building and deploying AI workloads.
Jaya Jagadish, Country Head and SVP Engineering, AMD India, said access to large-scale computing infrastructure, combined with the ROCm ecosystem, could help researchers, startups and enterprises develop AI applications with greater flexibility.
Growing demand for sovereign AI infrastructure
The deployment comes amid rising demand for local AI computing infrastructure as organisations seek to address data residency, regulatory compliance, security and performance requirements.
Vultr operates across 34 cloud data centre regions, serving enterprises, AI developers and government organisations in 17 countries across six continents.
The Chennai supercluster reflects the growing emphasis on distributing AI infrastructure across regions to bring computing resources closer to data sources and application users. For Indian organisations, the additional capacity is expected to provide another option for developing and running compute-intensive AI workloads on infrastructure located within the country.
