2 Sources
[1]
Neysa Deploys AMD GPU Cluster in Mumbai for AI Training and Inference
Neysa's Tytan platform allows for customisation of accelerator architectures based on specific workload needs. Neysa has deployed an AMD Instinct MI350X GPU cluster in Mumbai through its Tytan bare-metal AI infrastructure platform. The cluster is now operational and will support large language
[2]
Neysa deploys AMD's MI350X GPU cluster in Mumbai to expand AI compute capacity
Neysa has launched a large-scale AI cluster in Mumbai utilising AMD's Instinct MI350X GPUs for AI workloads. This deployment aims to enhance dedicated computing capacity for model training and large-scale inference. Customers can configure the infrastructure through options including bare metal,
Share
Copy Link
Neysa, an AI cloud provider in India, has brought an AMD Instinct MI350X GPU cluster online in Mumbai on its Tytan bare-metal platform, and the company says it is already live
1
2
. The addition gives India more dedicated AI compute capacity, per the company's own announcement.
Source: ET
Neysa co-founder and chief executive Sharad Sanghi says the question has moved beyond GPU counts: "The question for large AI deployments is no longer simply how many GPUs are available. It is whether the infrastructure can be deployed at scale, operated with certainty and configured for the workload"
1
.
Source: AIM
The setup pairs AMD's MI350X GPUs with EPYC processors and the ROCm software stack
1
2
. It is aimed at training and fine-tuning large language models, generative AI and high-throughput inference, as the company describes it1
2
.Sanghi's pitch is control: "dedicated bare-metal capacity and the freedom to choose the silicon, architecture and operating model that fit their requirements"
1
. Customers can choose among bare metal, bare metal with the Slurm scheduler, or bare metal with Kubernetes, and set their own network, storage and service-level requirements1
2
. Neysa also runs Velocis, a managed AI cloud service, and describes its approach as silicon-agnostic, letting customers pick accelerator vendors while Neysa handles deployment and operations1
2
.The Mumbai cluster is one piece of an expansion the company has been assembling all year, backed by up to $600 million in equity. Blackstone and co-investors agreed to provide it in February, with a further $600 million of debt planned, toward a goal of more than 20,000 GPUs across India
1
. In October, Neysa partnered with data centre operator ST Telemedia Global Data Centres to host up to 9,000 GPUs at the STT Noida 2 site1
. The first phase is expected in early 2027.Neysa is not the only company scaling AMD capacity in India. Vultr announced plans in October for a 10,000-GPU supercluster in Chennai that, if built as planned, would be India's largest AMD Instinct supercluster, with expansion planned in 2027
1
.Neysa says global customers have already contracted capacity for AI deployments in India
1
.Related Stories
Summarized by
Navi
21 Oct 2026•Technology

16 Feb 2026•Technology

20 Jul 2026•Technology
