Neysa Deploys AMD Instinct MI350X GPU Cluster in Mumbai
Neysa has brought an 8,192-GPU AMD Instinct MI350X cluster online in Mumbai on its Tytan bare-metal platform.
Key Takeaways
- Neysa has deployed an 8,192-GPU AMD Instinct MI350X cluster in Mumbai.
- The cluster runs on Neysa Tytan, the company's dedicated bare-metal AI infrastructure offering.
- Neysa said the cluster is now live.
- It uses AMD EPYC processors and the AMD ROCm open software ecosystem.
Table of Contents 7 sections
Neysa, an Indian AI cloud company, has deployed an 8,192-GPU AMD Instinct MI350X cluster in Mumbai on Neysa Tytan, its dedicated bare-metal AI infrastructure offering. The company said the cluster is now live and will support training, fine-tuning and large-scale inference workloads.
The cluster runs on AMD Instinct MI350X GPUs, AMD EPYC processors and the AMD ROCm open software ecosystem.
Neysa's Mumbai AMD cluster at a glance
| Detail | Value |
|---|---|
| Company | Neysa |
| Location | Mumbai |
| GPUs | 8,192 AMD Instinct MI350X GPUs |
| Other hardware and software | AMD EPYC processors, AMD ROCm open software ecosystem |
| Platform | Neysa Tytan, dedicated bare-metal AI infrastructure |
| Workloads | Large language model training, fine-tuning, generative AI and high-throughput inference |
| Deployment options | Bare metal, bare metal with Slurm, or bare metal with Kubernetes |
| Status | Live |
How Neysa Tytan lets customers pick their own chips
Neysa said the deployment expands its silicon-agnostic infrastructure strategy. Through Tytan, customers choose the accelerator architecture and operating setup for their workloads, while Neysa runs cluster deployment and operations. Customers can deploy Tytan as bare metal, bare metal with Slurm, or bare metal with Kubernetes.
Neysa also runs Velocis, a managed AI cloud platform that brings together GPU compute, orchestration, inference, observability and cost management. The company is backed by Blackstone.
| When | What happened |
|---|---|
| February 2026 | Blackstone and co-investors agreed to provide up to $600 million in equity to Neysa, which also planned to raise $600 million in debt |
| October 2026 | Neysa picked STT GDC's Noida 2 data centre for a first phase supporting up to 9,000 GPUs |
| November 2026 | Neysa's AMD Instinct MI350X cluster in Mumbai went live |
Sharad Sanghi on choice of silicon for large AI deployments
"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."
"Neysa Tytan gives AI offtakers that control: dedicated bare-metal capacity and the freedom to choose the silicon, architecture and operating model that fit their requirements. With AMD Instinct MI350X on Tytan, we are expanding that choice while delivering the scale global customers need to deploy AI infrastructure in India."
AMD's India sales head on the Neysa collaboration
"Through our collaboration with Neysa, we are expanding access to advanced AI infrastructure in India and supporting the development of a more diverse and resilient AI ecosystem."
Neysa's next GPU site in Noida
In October 2026, Neysa said it will deploy GPU infrastructure and its Velocis platform at STT GDC's Noida 2 data centre. STT GDC said the first phase will support up to 9,000 GPUs and is expected to go live in the first quarter of 2027.
Where Neysa stands after the Mumbai launch
Neysa now has an 8,192-GPU AMD Instinct MI350X cluster live in Mumbai. STT GDC said the first phase of Neysa's deployment at its Noida 2 data centre will support up to 9,000 GPUs and is expected to go live in the first quarter of 2027.
More news
Frequently Asked Questions
What is Neysa?
What did Neysa announce?
How many GPUs are in Neysa's Mumbai cluster?
Where is Neysa's new AMD cluster located?
How was Neysa's Mumbai cluster brought online?
What workloads will Neysa's AMD cluster support?
What hardware and software power Neysa's cluster?
How can customers deploy Neysa Tytan?
Who leads Neysa?
Who is backing Neysa?
How many GPUs will Neysa's Noida site support?
More in AI
Velocity Launches Vector, an AI Agent Platform for D2C Shipping
Velocity's new Vector platform uses AI agents to confirm orders, pick couriers and handle failed deliveries.
HCLTech and Cohere Partner to Build Sovereign AI
HCLTech and Cohere will build sovereign AI on Cohere's North platform, starting in Canada and Germany.
KiteFishAI Launches NanoI1, a 155M Open-Weight AI Model
KiteFishAI has launched NanoI1, a compact open-weight AI model built for specialised enterprise tasks.
