Data centers · Oct 21, 2026, 16:44
Vultr to Deploy 20 MW AMD MI355X Cluster in Chennai, India's Largest
Vultr plans a 20 MW supercluster of AMD Instinct MI355X GPUs in Chennai, with 288 GB of HBM3E per GPU, but has not said the GPU count, host site or start date.
AMD's official image of an Instinct MI350X accelerator; the Chennai cluster will use the related MI355X, which this photo does not show.
Image: AMD
Source · Press image from the AMD website, used for news reporting.
Vultr said it will deploy AMD Instinct MI355X GPUs in a new 20 MW supercluster in Chennai, India, which the cloud provider calls the country's largest AMD Instinct deployment, according to Analytics India Magazine and the company's announcement. The project would give Vultr two large AMD clusters, alongside its supercluster in Springfield, Ohio.
Each MI355X in the Chennai cluster carries 288 GB of memory, against 180 GB on Nvidia's B200 as cited by Analytics India Magazine.
Illustration (AI-generated)
Each MI355X carries 288 GB of HBM3E memory with 8 TB/s of memory bandwidth, Vultr said. Analytics India Magazine compares that with 180 GB on Nvidia's B200 and notes that Nvidia keeps an advantage through its CUDA software ecosystem. The Chennai cluster is aimed at AI labs, foundation-model builders, startups, enterprises and research institutions for training, fine-tuning and inference. Vultr CEO J.J. Kardwell framed it around data sovereignty, data residency and latency requirements, saying organizations in India need in-country infrastructure.
Sources
- Vultr to deploy 20MW AMD MI355X cluster in Chennai, India · DatacenterDynamics