AI Cloud Services

Infrastructure for your AI needs

Our AI Cloud Services are designed to scale according to your training and inference needs while ensuring smooth data management.

Bare metal AI clusters

Dedicated bare metal GPUs available in multiples of 8 GPUs per node.

GPU: 8*H100 80GB SXM5
CPU: Intel Xeon Platinum 32 Core (128vCPU), Dual Socket
Network: ConnectX-7 NDR Blue Field-3 DPU
Memory: 2048GB DDR5
4800MT/s RDIMMs
Storage: 62TB NVMe inbuilt storage + Additional storage
as needed
GPU Interconnect: NVLink high throughput interconnect

Virtual machines (VMs)

Ready-to-use, flexible VMs powered by NVIDIA GPUs.

GPU: Starting from a single GPU
CPU: Starting from
2vCPU per VM
Network: ConnectX-7 NDR Blue Field-3 DPU
Memory: Starting from
4GB per VM
Storage: Starting from
10GB per VM

Flexible storage options

High-performance Tier 1 or Tier 2 solutions to suit your specific storage, retrieval, and workload requirements.

High-performance Tier 1 or Tier 2 storage catering to every need of AI apps
Up to 1Gpbs read performance
per GPU
700 Mps write performance
per GPU
Available in flexible sizes
Option to be coupled with bare metal AI clusters and VMs

Discover how AI
transforms industries

Benefits

Ease of management

Our unified multi-cloud, multi-access orchestration Paragon™ platform provides unparalleled automation, visibility, and control that cannot be found anywhere else.

Scalability and efficiency

Fine-grained control and flexible service options provide cost-efficient, scalable infrastructure configurations for all needs and scenarios.

Accelerate time to market

Models-as-a-service and integration with repositories like HuggingFace speed up model development and training.

Features

Paragon™ management plane

Experience effortless infrastructure management with Paragon™, Singtel’s patented platform that revolutionises AI cloud orchestration.
Recognised for its award-winning capabilities, Paragon™ provides seamless multi-cloud and multi-access orchestration, automating lifecycle management across the board.
With its single-pane-of-glass interface, you gain complete visibility and control over your AI infrastructure, from GPUs to edge devices – ensuring efficiency and agility at every level.

Turnkey AI workspaces

Automated setup of AI workspaces, tailored to the requirements of data scientists and AI engineers, lets you begin training and tuning your models without the hassle of infrastructure DevOps.

Models as a service

RE:AI Cloud workspaces are pre-loaded with industry-leading open-source models, available through a localised repository, to accelerate model development. Host and manage your own custom models trained on top of existing base models.