Deployment Timeline

Progressive delivery of production-ready GPU cloud capacity across 4 phases.

Deployment Overview

Key metrics summarizing the scale and scope of the GPU cloud deployment program.

128

GPUs (Phase 1)

512

GPUs (Phase 2)

1,024

GPUs (Phase 3)

2,048+

GPUs (Phase 4)

18

Months (Total Timeline)

4+

Regions (APAC Coverage)

Phase Timeline

Detailed breakdown of each deployment phase, milestones, and current status.

Phase 1 — Month 1-3: Foundation

Foundation

Data center site preparation
Network deployment (spine-leaf, InfiniBand)
First GPU rack: 16 x 8-GPU nodes (128 x H100 SXM5 80GB)
Platform software deployment
Pilot program with 5-10 enterprise clients

Phase 2 — Month 4-6: Public Launch

Public Launch

Additional 384 GPUs (H100 + B200 mix)
Serverless GPU endpoint launch
Instant Clusters for distributed training
Template marketplace launch
API and SDK release

Phase 3 — Month 7-12: Scale & Enterprise

Scale & Enterprise

GPU expansion to 1,024 units
Multi-region deployment (Jakarta + Singapore)
Enterprise features: VPC, dedicated tenancy
SI services launch

Phase 4 — Month 13-18: APAC Expansion

APAC Expansion

GPU fleet to 2,048+ units
Additional regions: Tokyo, Sydney, Mumbai
Partner SI program

GPU Hardware Specifications

Technical specifications for GPU accelerators deployed across the cloud platform.

GPU Model	Memory	Interconnect	TDP	Primary Use
NVIDIA H100 SXM5	80 GB HBM3	NVLink 4th Gen	700W	Training, Inference
NVIDIA B200	192 GB HBM3e	NVLink 5th Gen	1000W	Next-gen Training
NVIDIA A100 SXM4	80 GB HBM2e	NVLink 3rd Gen	400W	Cost-effective Training
NVIDIA L40S	48 GB GDDR6	PCIe Gen5 x16	350W	Inference, Rendering