Single architecture spanning edge, on-premises, managed compute, and private AI cloud.
CoralDC Platform
One Infrastructure Architecture. From Edge to Enterprise.
CoralDC unifies AI Factories, Private AI Cloud, GPU Infrastructure, Edge AI, managed compute, and immersion-cooled systems into a deployable infrastructure architecture for organisations that need to govern AI under their own control.
Infrastructure Architecture
Sovereign AI infrastructure system
Immersion-cooled compute for governed AI environments.
Product Architecture
Coral Edge and Coral OnPrem starting points.
Coral Edge 2UCoralDC Coral Edge 2U visualCoralDC infrastructure visual
Coral Edge 2U
Compact AI infrastructure for branch, facility, and micro-edge deployment
Coral Edge 4UCoralDC Coral Edge 4U visualCoralDC infrastructure visual
Coral Edge 4U
High-density edge compute for multi-tenant and GPU-heavy environments
Coral Edge 6UCoralDC Coral Edge 6U visualCoralDC infrastructure visual
Coral Edge 6U
Micro-pod AI infrastructure for ruggedised and mission-critical edge environments
Coral OnPrem 12UCoralDC Pod12 visualCoralDC infrastructure visual
Coral OnPrem 12U
Departmental AI, HPC, and private compute in a compact immersion platform
Coral OnPrem 24UCoralDC Pod24 visualCoralDC infrastructure visual
Coral OnPrem 24U
Enterprise-scale private cloud and high-density compute consolidation
Coral OnPrem 42UCoralDC Pod42 visualCoralDC infrastructure visual
Coral OnPrem 42U
Full-rack immersion infrastructure for AI factories and sovereign compute
| Model | Positioning | Approx. IT Load | Compute Profile | Best Fit |
|---|---|---|---|---|
| Coral Edge 2U | Compact AI infrastructure for branch, facility, and micro-edge deployment | Approx. 4.6 kW | 1-2 dual-socket servers; up to 4 GPUs; up to approximately 100 TB NVMe where configured | Retail micro data centres, industrial IoT gateways, telemedicine edge, video analytics, compact inference |
| Coral Edge 4U | High-density edge compute for multi-tenant and GPU-heavy environments | Approx. 9.2 kW | 2-4 dual-socket servers; up to 8 GPUs; up to approximately 250 TB NVMe/SAS storage | Telco MEC, smart campuses, autonomous logistics, hospital edge PACS, regional inference nodes |
| Coral Edge 6U | Micro-pod AI infrastructure for ruggedised and mission-critical edge environments | Approx. 13.8 kW | 4-6 compute nodes; GPU or accelerator options; approximately 400-500 TB high-performance storage | Mining, energy, 5G/ORAN hubs, smart manufacturing cells, critical infrastructure nodes, mobile compute |
| Model | Positioning | IT Load Cooling Capacity | Compute Profile | Best Fit |
|---|---|---|---|---|
| Coral OnPrem 12U | Departmental AI, HPC, and private compute in a compact immersion platform | 15-40 kW | Up to 12 x 1U dual-socket servers; optional hot-swap trays | Departmental HPC, private AI clusters, quant/risk engines, CAD/PLM, research compute, constrained data centres |
| Coral OnPrem 24U | Enterprise-scale private cloud and high-density compute consolidation | 50-100 kW | Up to 24 x 1U dual-socket servers; dense CPU nodes; mixed CPU/GPU configurations | Private cloud clusters, database farms, VDI, analytics, telco core, enterprise AI platforms |
| Coral OnPrem 42U | Full-rack immersion infrastructure for AI factories and sovereign compute | 100-200 kW | Up to 42 x 1U dual-socket servers; high-end GPUs/accelerators; up to approximately 500 TB storage where configured | AI training clusters, model serving, national-scale private cloud, rendering, trading, telco modernisation, government AI factories |
| Buyer Type | Recommended Starting Point | Rationale |
|---|---|---|
| Edge AI Leader | Coral Edge 2U, 4U, or 6U | Deploy local inference, analytics, and storage close to operations. |
| Enterprise AI Leader | Coral OnPrem 12U or 24U | Establish private AI capacity, consolidate workloads, and build repeatable infrastructure capability. |
| Government / Research Leader | Coral OnPrem 24U or 42U + Coral Cloud | Create sovereign AI infrastructure with governance, orchestration, and scalable GPU capacity. |
| Industrial Operator | Coral Edge 6U + Coral OnPrem regional node | Support remote inference while centralizing model training and operational analytics. |
FAQ
Direct answers for infrastructure buyers.
What is the CoralDC Platform?
It is the unified architecture connecting Coral OnPrem, Coral Cloud, Coral Compute, Coral Edge, and immersion cooling.
Is this only hardware?
No. CoralDC positions infrastructure, cooling, governance, deployment, and operational architecture as one system.
Where should buyers start?
Edge leaders should start with Coral Edge; enterprise and sovereign buyers should start with Coral OnPrem, Coral Cloud, and an architecture review.
What is the difference between Coral Edge and Coral OnPrem?
Coral Edge is built for distributed sites: smaller deployments where proximity to data, poor connectivity, harsh conditions or a sealed environment make a conventional facility impractical. Coral OnPrem is built for higher-density private AI inside an institution's own premises, typically a larger deployment supporting research computing, training or sustained inference.
Does a Coral deployment need a data centre?
No. That is the central design constraint. Coral systems are self-contained and liquid-cooled, so they do not require a raised floor, computer room air handling, or a purpose-built hall. They install in existing rooms, subject to floor loading, electrical capacity and access assessment during the architecture review.
Can CoralDC operate the infrastructure for us?
Yes. Coral Compute is managed GPU infrastructure and operations, designed for organisations that need AI capacity without building and retaining a data centre engineering team. The infrastructure remains on the customer's premises and under the customer's jurisdiction.
The Coral Platform
What is actually in the platform
CoralDC builds self-contained, liquid-cooled AI infrastructure that installs inside existing buildings. No data hall, no raised floor, no air-handling retrofit. The platform spans distributed edge deployments through high-density on-premises systems, with managed GPU operations and private AI cloud orchestration running on top of the same hardware.
Coral Edge
Distributed AI infrastructure for sites where latency, connectivity, physical conditions or jurisdiction rule out a conventional facility. Sealed, liquid-cooled, designed to operate in rooms that were never built for compute. …more
Coral OnPrem
High-density private AI infrastructure for enterprises, governments, health systems and research institutions operating under their own control. …more
Coral Compute
Managed GPU infrastructure and operations, so a customer can run AI workloads without building a data centre team. …more
Coral Cloud
Private AI cloud orchestration across CoralDC and customer-controlled environments, with jurisdictional control retained by the institution. …more
One Architecture
Why the platform is built as one architecture
A conventional AI deployment asks the buyer to solve power, cooling, water, connectivity, compliance and operations separately, each with a different vendor, a different timeline and a different point of failure. The integration risk sits entirely with the customer.
Coral's platform is designed so those decisions arrive already made and already tested together. One thermal architecture selected for the customer's actual hardware generation. One power and grid interface. One telemetry model. One operations layer. One commercial agreement.
Boundaries
What the platform does not do
CoralDC integrates specialist components rather than manufacturing GPUs, fluids or generation equipment. Its scope is pod-level thermal, hydraulic and controls integration, together with telemetry and operations.
Internal Links
Continue through the architecture.
Next Step
Explore the CoralDC Platform
Every CoralDC engagement should connect executive intent, technical architecture, economics, and deployment reality before infrastructure decisions are made.