CHENNAI: US-based cloud infrastructure provider Vultr on Monday launched a 20-MW AI supercluster in Chennai, powered by AMD Instinct MI355X GPUs, marking its largest such deployment in India.
The facility, expected to house around 10,000 GPUs in its initial configuration, will be among Vultr’s two largest AMD AI infrastructure deployments globally.
Designed to support large language model and foundation model training, the facility will cater to AI labs, startups, enterprises, research institutions and universities. Kevin Cochrane, chief marketing officer, Vultr, said, “India is our largest market outside the US and it has highest footprint across the globe.”
Vultr plans to expand capacity in 2027 with AMD’s MI455X GPUs and Helios rack-scale AI architecture. The company is targeting 600 MW of AI infrastructure capacity in India by 2028 and gigawatt scale by 2029, although it has not committed to firm timelines.
Vultr did not disclose the investment involved in the Chennai project or its existing data centre operations in Delhi-NCR, Mumbai and Bengaluru. Its Chennai workforce could eventually reach 5,000 as the facility becomes operational and services expand.