NVIDIA B300

Enterprise GPU Hardware. No Compromise.

Singapore's first dedicated B300 compute facility. Direct access to current-generation GPU infrastructure, available from Q4 2026.

The Platform

What Direct Access to B300 Means in Practice

You get dedicated capacity on physical hardware — no shared resource pool, no performance overhead from multi-tenancy. Your workload runs on isolated B300 GPUs in a facility purpose-built for their thermal and power requirements. Bare metal access is available as standard for operators who need it.

The operator relationship is direct. So is the hardware access.

Cluster Architecture

Built for Scale. Configured for Performance.

Each compute cluster is structured from the GPU up — nodes grouped into racks, racks into clusters — with power and cooling infrastructure matched to the demands of high-density GPU operation. Flexible access models from direct hardware access to fully managed orchestration. Full topology is available to qualified customers.

Key Specifications

SIN1 — B300 Cluster

GPU Hardware
NVIDIA B300
Total GPU Count
~736
Memory per GPU
288GB HBM3e
Cooling
Direct liquid cooling
Availability
Q4 2026
Access Model
Bare metal or managed (Slurm, Kubernetes)

Looking Ahead

Designed for What's Next

SIN1 is designed to accommodate next-generation GPU architectures as they become available. We are actively evaluating upcoming platform generations and our infrastructure is built to support the thermal and power profiles they demand. Speak to us about forward capacity planning.

Ready to Go Deeper?

Full cluster topology, node-level specifications, and capacity terms can be discussed with qualified customers. Start with a conversation.