Singapore Colo Node 01
Colocation / enterprise edge · Singapore, Singapore- Power
- 24.2 kW
- Headroom
- 8.4 kW
- Latency
- 12 ms
- Uptime
- 99.96%
APAC fintech and enterprise AI inference
Commercial sites become governed AI infrastructure.
v5 adds key rotation and revocation, GPU telemetry collection, heartbeat tracking, secure ingestion, and explainable routing foundations.
online
Dar Tower Node 01
Telecom tower · Dar es Salaam, Tanzania- Power
- 18.4 kW
- Headroom
- 5.6 kW
- Latency
- 21 ms
- Uptime
- 99.92%
Swahili voice and tourism inference
Frankfurt Sovereign Node
Enterprise / sovereign AI site · Frankfurt, Germany- Power
- 30.8 kW
- Headroom
- 9.2 kW
- Latency
- 18 ms
- Uptime
- 99.88%
EU data-residency inference
Dubai Enterprise Campus Node
Enterprise campus · Dubai, United Arab Emirates- Power
- 28.7 kW
- Headroom
- 3.1 kW
- Latency
- 44 ms
- Uptime
- 98.74%
Arabic document AI and customer service
Ashburn Edge Node
Regional data-center room · Ashburn, United States- Power
- 0 kW
- Headroom
- 0 kW
- Latency
- — ms
- Uptime
- 97.31%
Pending maintenance window
Onboarding flow
From host site to production inference.
Qualify site
Confirm power capacity, cooling path, physical security, internet failover, and local permit requirements.
Create site record
Register the host organization, site, power profile, support contact, and risk score.
Install node agent
Deploy the UmamiEdge node agent to send GPU, power, thermal, and network telemetry from nvidia-smi.
Approve production routing
Run soak tests, bind the node to a tenant group, then enable inference traffic with SLA guardrails.