Vultr has launched a 20-megawatt AI supercompute cluster in Chennai. It is the cloud provider’s second-largest globally, behind only its flagship facility in the United States, and a significant new option for enterprises that want AI capacity inside India.
The cluster is built on more than 10,000 AMD Instinct MI355X GPUs and AMD EPYC CPUs, and is aimed at large-scale AI training and inference. It targets AI labs, startups, enterprises and research institutions that need in-country compute, driven by demand for locally hosted workloads that meet data residency, sovereignty and latency requirements.
For CIOs, the headline is less about megawatts than about where AI workloads can run. Training, fine-tuning and inference on Indian soil are increasingly a compliance and performance question, not only a cost one. Vultr’s CEO framed the cluster as giving Indian organisations the capacity to train, fine-tune and run inference on in-country infrastructure.
The software stack matters too. The cluster will use AMD’s open ROCm ecosystem, so teams with CUDA-heavy pipelines should factor in migration effort and testing before moving workloads.
gktoday
A fast build-out
Vultr entered India in 2022 with three regions: New Delhi NCR, Bengaluru and Mumbai. Those initially ran on AMD EPYC processors. The company has since moved up the AMD GPU line, from the MI300X to the MI325X and now the MI355X.
Vultr calls itself the world’s largest privately held independent cloud infrastructure provider, with a footprint across six continents. Its network is designed to connect up to 90% of the world’s population within two milliseconds.
The roadmap goes further. Vultr plans significant additional capacity in 2027, including AMD Instinct MI455X GPUs and the AMD Helios rack-scale architecture. One report says the company is aiming for 600 megawatts of AI capacity in India by 2028 and gigawatt scale by 2029. That figure comes from a single secondary source, so treat it as unconfirmed.
Tamil Nadu moves to put startups on the cluster
Alongside the launch, Vultr India signed an MoU with StartupTN, the Tamil Nadu Startup and Innovation Mission. It extends cloud credits and preferential pricing to eligible startups across the state. Early-stage startups get US$2,500 in Vultr credits along with a 20% discount, while eligible startups beyond the early-stage funding threshold get US$45,000 in credits along with the same 20% discount. The aim is to cut infrastructure costs for startups building and scaling technology products, particularly in AI and deep tech.
Kevin Cochrane, Chief Marketing Officer, Vultr, pointed to India’s shift from consuming AI to building it, and to its engineering talent as a driver of the investment. He said Vultr is strengthening its developer relations function and supporting India’s open-source developer ecosystem. “India isn’t just consuming AI. The buildout of AI is happening here. “This is where the talent is. This is where the challenge is literally changing the world.”
Thiru. P Mathan Raja, Minister of Micro, Small & Medium Enterprises, Government of Tamil Nadu, stressed access for startups and MSMEs, including those in smaller cities. He cited uses such as product design, software testing, inventory management and operational efficiency.
“The true measure of technology is not how sophisticated it is, but the opportunity it creates.” Thiru. Dr. Atul Anand, IAS, Additional Chief Secretary, IT & Digital Services Department, cited Tamil Nadu’s engineering talent and its roughly 16,000 startups. He argued that affordable compute is essential to accelerating AI innovation, urged Vultr to extend accessible pricing to Tamil Nadu startups, and assured the company of state government support.
Thiru Singai G. Ramachandran, Chairman, Tamil Nadu Startup & Innovation Mission, highlighted the state’s manufacturing, electronics and deep-tech base. He cited Chennai’s approximately 567 deep-tech startups, with an estimated ecosystem valuation of ₹75,000 crore, and the need to build AI and software capabilities alongside manufacturing strengths. His message to the industry was about outcomes, not capacity: “Beyond measuring infrastructure by megawatts, measuring its impact is very important.”
The Chennai cluster won’t settle India’s compute question by itself. But with capacity close to the developers and enterprises who need it, state-backed startup access and a clear AMD roadmap behind it, it gives technology leaders a real option to weigh as AI workloads move closer to home.