> Fabric for AI
Accelerate AI workloads from end to end
Transform data into intelligence faster with seamless connectivity across clusters, models, workflows, and environments.
Compute + data movement drive AI success
Traditional networks can’t catch up to dynamic AI workloads that need rapid iterations, flexibility, visibility, and real time control. As AI models scale across regions and environments, the network is no longer just a transport layer. It’s become a core AI lifecycle orchestrator, enabling low latency movement of data, models, and GPU clusters to accelerate optimization end to end.
Built for AI data, models, and compute
Optimize AI connectivity with an ultra-low latency global fabric that links compute across all your clouds and regions.
Optimize AI operations worldwide
Automate orchestration
Configure connections instantly through APIs that integrate directly with orchestration systems.
Scale on demand
Establish high-bandwidth paths in minutes during training without deploying any hardware
Connect to clouds faster
Ingest training data rapidly with direct connections to AWS, Azure, Google Cloud, and more
Keep data secure
Move models and training data on dedicated, private Layer 2 links free from prying eyes and jitter.
Accelerate edge inference
Run inference at edge locations near your users for faster responses and lower latency.
Interconnect across regions
Link major AI hubs like Singapore, Hong Kong, Tokyo, Jakarta, Los Angeles, and Frankfurt.
Singapore metropolitan network
Connect directly to Asia's core compute clusters with high-capacity bandwidth and low latency.
Data center overview
Core data centers
- Global Switch TaiSeng
- Equinix SG1
Edge data centers
- Global Switch Woodlands
- Equinix SG2
- Equinix SG5
- Telin 3 Data Center
- DRT Jurong
- Keppel DC SGP1
- Equinix JH1
Networking capabilities
Network redundancy
- Multiple route options provided
- KMZ files available
Network latency
- Less than 1 ms
Ports
- Supports 10GE or 100GE
- 400GE ports available in core locations on request
Underlying technology
- DWDM
Singapore-Johor dedicated line
Move AI data with ultra-low latency between Asia’s key AI hubs.
Network Capacity
- Equinix JH1 – Equinix SG1: 2 x 1.6 T
- Equinix JH1 – Global Switch: 2 x 1.6 T
- Equinix SG1 – Global Switch: 2 x 1.6 T
Network Latency
- Less than 2 ms
Ports
- Supports 10GE or 100GE
- 400GE ports available on demand
Scale AI without boundaries
Connect with an AI expert to power your AI workloads globally with reliable, ultra-low latency network performance.
Global service, local support
24/7 live technical support included
< 15 minute
response time
