US-based cloud infrastructure company Vultr on 2 November 2026 launched a 20-megawatt AI supercluster in Chennai, built on AMD Instinct MI355X GPUs. It is the company’s largest AI deployment in India and is expected to hold around 10,000 GPUs in its first phase.

Vultr said the Chennai site will rank alongside its Springfield, Ohio facility as one of its two biggest AMD AI infrastructure builds anywhere in the world. The company has set an ambition of 600 MW of AI capacity in India by 2028 and gigawatt scale by 2029, though it has not committed to firm timelines.

What will the Chennai AI supercluster do?

The cluster is designed for training and fine-tuning large language models and foundation models, and for high-throughput inference. Vultr is pitching it at AI labs, startups, enterprises, research institutions and universities.

• Hardware: AMD Instinct MI355X GPUs, roughly 10,000 in the initial configuration.
• Software: AMD’s open ROCm ecosystem, giving developers access to common AI frameworks, libraries and models.
• Next phase: an expansion in 2027 using AMD MI455X GPUs and the Helios rack-scale architecture.
• Power: the Tamil Nadu government has enabled 150 MW of connected power at the site, with 20 MW available for the first deployment.

Capacity already spoken for

Chief Marketing Officer Kevin Cochrane told The Hindu that global enterprises have pre-reserved a substantial share of the Chennai capacity, without naming them or the volume booked. Part of the capacity will still be made available to digital startups and Indian universities, he said, and the company’s expansion over the next two to three years will be “an order of magnitude greater” than this first build.

“I would be disappointed if we don’t have 600 megawatt live by 2028, and then get into gigawatt by 2029,” said Amit Rai, General Manager, AI and Enterprise Cloud, Vultr.

Vultr did not disclose the investment in the Chennai project or in its existing data centres in Delhi-NCR, Mumbai and Bengaluru. It also did not share the facility’s cooling technology, water use or energy-efficiency figures, saying it is working with real estate developers, data centre operators and server maker Supermicro on efficient racks.

Why it matters for Indian CIOs

Access to GPUs inside India has become a deciding factor for enterprises that must keep sensitive data in the country. The government is adding public compute through the IndiaAI Mission’s 10,000-GPU tender, while private providers are racing to turn data centre space into usable AI capacity, a gap highlighted in a recent Rhodium Group report on India’s AI data centre boom. Domestic players are also moving up the stack, as seen with E2E Networks’ pay-per-token Blaze inference service.

Vultr, which works with both AMD and NVIDIA, is positioning its multi-vendor approach as a way for customers to choose their GPU stack. Its leadership claimed savings of 50 to 90 per cent on core cloud computing compared with traditional hyperscalers, but did not quantify savings on GPU workloads.

Hiring plans

Vultr has more than 10,000 active customers in India and has hired over 100 people in the country in the past three months. It plans thousands more hires in network engineering and operations by March, and said its Chennai workforce could eventually reach 5,000 as the facility comes online.

Key takeaways

• Vultr has launched a 20 MW AMD-powered AI supercluster in Chennai with about 10,000 MI355X GPUs.
• Global enterprises have pre-booked a large part of the capacity; startups and universities will also get access.
• The company is targeting 600 MW in India by 2028 and 1 GW by 2029, without firm commitments.
• Investment size, cooling and water-use details were not disclosed.

Be a part of Elets Collaborative Initiatives. Join Us for Upcoming Events and explore business opportunities. Like us on Facebook , connect with us on LinkedIn and follow us on Twitter.

"Exciting news! Elets technomedia is now on WhatsApp Channels Subscribe today by clicking the link and stay updated with the latest insights!" Click here!