01 Overview

What is Eminence Sovereign?

Eminence Sovereign is Eminence Grey’s distributed sovereign AI infrastructure platform. It integrates private compute, networking, security, and operations through Whitehorse, with an operating scope defined by one service agreement.

02 Architecture

One architecture, from fabric to edge

Private connectivity links approved environments and compute. Whitehorse coordinates control across infrastructure layers. Whitehorse Outrider establishes the customer-premises boundary.

Customer applications remain above the infrastructure layer; their compatibility and integration are evaluated during scoping.

03 Capabilities

Capabilities within the platform

Whitehorse

Distributed orchestration, policy, workload lifecycle, and evidence.

Post-Quantum Security

Defined protection, key handling, and implementation boundaries.

04 Compute

Match infrastructure to the workload

Evaluate reasoning, inference, private training, and HPC against compute, connectivity, tenancy, and control requirements. Existing models and open-weight models are assessed for technical and licensing fit.

The proposed catalog includes NVIDIA DGX B200 and H200 systems, Grace-based configurations, and EPYC-based HPC options. NVIDIA AI Enterprise applies where specified. Lumen provides the private optical-backbone component.

Availability, licensing, reach, capacity, latency, and permitted data movement are confirmed for the deployment.

05 Edge

Whitehorse Outrider

Outrider brokers approved traffic between customer premises and the fabric. Inspection, access, and policy are defined at that boundary, without assuming every workload must move.

Government deployments may combine Outrider with Mobile or Edge Nodes.

06 Facilities

Configure the environment

The facility catalog spans Alpha+, Bravo+, Charlie+, Delta+, Delta-Lite, Echo, and Foxtrot-Mix. Select by workload, location, tenancy, capacity, and controls.

Delta+, Delta-Lite, and defense-tier Echo support defense scoping. Foxtrot-Mix is a regulated multi-tenant configuration; it is not equivalent to dedicated tenancy.

07 Comparison

Compare what you must operate

Colocation

What must you procure and operate beyond facility services?

Cloud / neocloud

Which tenancy, connectivity, placement, and shared responsibilities apply?

Private / sovereign cloud

Which technical and governance boundaries are included?

Application / mission software

How will it integrate with the infrastructure?

08 Deployment

Define the deployment

Continuity planning addresses workload placement, recovery capacity, connectivity, and testing.

Requirement

Start with the workload: the information and models that matter, and the controls and evidence it must meet.

Topology

Place compute, connectivity, and the customer-premises boundary around where information may run and how it may move.

Compute

Match compute, tenancy, and location to reasoning, inference, private training, or HPC.

Access

Define who can enter the environment, under which policies, and how approved traffic crosses the boundary.

Controls

Map applicable requirements to the actual system boundary, assessment scope, and responsibilities.

Acceptance evidence

Validate the deployment against agreed acceptance criteria and the evidence your security and audit teams need.

09 Agreement

One agreement. A defined scope.

The service model combines fabric, compute, control, and operations. The agreement defines consumption terms, responsibilities, evidence, and separately scoped additions.