Original NVIDIA hardware
DGX Spark Founders Edition with the original NVIDIA GB10 superchip.
128 GB unified memory20-core Arm CPUUp to 1 PFLOP FP4
NVIDIA DGX Spark Founders Edition+Galene.AI Elettra
Production-ready private and sovereign AI appliance
Private and Sovereign AI. Ready on your desk.
Galene.AI Elettra turns NVIDIA DGX Spark from compact AI compute into a production-ready AI appliance for professionals and teams that want cloud-like capabilities on their desk, with total control over data, models, and runtime.
Private and sovereign by design
Data, models, and runtime controls stay inside your own perimeter.
Zero data egressLocal inferencePolicy enforcement
Product, not project
Galene.AI Elettra turns NVIDIA AI compute built for developers into a production-ready private and sovereign appliance.
Cloud-like UXAgentic workflowsOps-ready
Why this appliance
AI should run where your data already lives.
Cloud AI is easy to start, but pay-per-use economics quickly become a burden. When every prompt, workflow, or experiment raises a cost question, creativity slows down and AI adoption hits a limit.
Galene.AI Elettra combines compact NVIDIA compute with the Galene.AI platform layer, so cloud-like AI capabilities can run locally on a sovereign desk-side machine. You buy the appliance once, use it for years, and get 3 years of software updates and technical maintenance included in the price.
Pay once3 years includedZero data egressCloud-like UX
Product, not project
DGX Spark is the hardware. Galene.AI Elettra is the product.
A complete turn-key appliance, ready to use out of the box, without stitching together models, connectors, governance, and operations yourself.
Turn-key out of the boxNo integration projectProduction-ready interfaces
Sovereignty by design
Private and sovereign, by design.
Data, permissions, models, and runtime controls stay inside your perimeter. Generative Shield adds GDPR and AI Act compliance, auditability, and full platform observability.
Generative ShieldGDPR and AI ActFull observability
One lifecycle
One maintained product, not a stitched stack.
Instead of designing, securing, updating, and maintaining your own agentic AI stack, teams get one software product with clear updates, maintenance, and operational ownership.
Single software productClear updates and maintenanceLess stack to maintain
AI adoption journey
From private chat to integrated AI workflows.
Start immediately with private chat grounded on your documents. Then build specialized agents on your data. Finally, integrate Galene.AI Elettra with company tools and systems through workflow automation and APIs.
Start now
01Start with private chat on your documents
Launch a private assistant that works on your files and internal knowledge from day one, without sending data or prompts outside your perimeter.
- Private chat
- Document-grounded answers
- Immediate time to value
Specialize
02Build specialized agents on your data
Move beyond chat with focused agents that work on your knowledge, tools, and internal procedures while keeping permissions and runtime under control.
- Specialized agents
- Controlled permissions
- Data-aware workflows
Integrate
03Integrate with tools and systems
Use workflow automation and APIs to embed Galene.AI Elettra into company tools, business systems, and operational processes at a deeper level of integration.
- Workflow automation
- API integration
- Company systems
Galene.AI Elettra
AI Operating System.
Galene.AI Elettra brings the software product layer that turns DGX Spark into a usable private and sovereign appliance: assistants, agents, workflow automation, APIs, integrations, and controlled rollout in one operating layer.
Assistant
Private copilot
General-purpose AI Assistant
A secure internal assistant for everyday work, grounded on enterprise knowledge inside your sovereign perimeter.
Private chatKnowledge groundingDaily copilot
AI Adoption
- Secure internal assistant
- Knowledge grounding
- In-perimeter usage
Scope
- Daily productivity
- Internal research
- Enterprise Q&A
Agents
Operational agents
No-code AI Agents Builder
Build and run assistants and specialized agents without writing code, with permissions, tools, and runtime under control.
Tool permissionsTask chainingRuntime guardrails
AI Adoption
- No-code construction
- Controlled permissions
- Clear execution patterns
Scope
- Specialized agents
- Repeatable tasks
- Department workflows
Automation
Live workflows
Private Workflow Automation Hub
Orchestrate end-to-end workflows inside your perimeter, with AI actions connected to real systems and business processes without giving up control.
System actionsWorkflow runsBusiness triggers
AI Adoption
- End-to-end orchestration
- Actions on real systems
- Private execution paths
Scope
- Internal processes
- Operations
- Business automation
APIs
Product interfaces
OpenAI-compatible SDK and REST APIs
Integrate Galene.AI Elettra into applications and products using familiar interfaces without giving up sovereignty, control, or ownership.
OpenAI-compatibleTool callingEmbeddable APIs
AI Adoption
- Compatible interfaces
- Structured tool calling
- Product integration ready
Scope
- Custom apps
- Digital products
- Dedicated front ends
CLI
Private coding
Private CLI for coding and console automation
Use local models for coding assistance and console automation inside your own perimeter, as a private alternative to cloud-based coding copilots.
Coding assistanceConsole automationLocal models
AI Adoption
- Private development support
- Terminal actions
- Local execution
Scope
- Engineering workflows
- Console tasks
- Private technical operations
Enterprise integrations
Connected to the systems your teams already rely on.
Galene.AI Elettra carries the same Galene.AI integration logic, so assistants, agents, and workflows can work across identity, knowledge bases, storage, databases, business systems, and MCP servers through a cloud-like product layer that stays private, sovereign, and under operational control.
Generative Shield
Runtime governance and risk mitigation.
Standard and custom controls applied at runtime, with approvals, traceability, and secure operations inside your perimeter.
Generative Shield is Galene.AI's proprietary runtime layer for risk mitigation and governance. It applies policy controls across assistants, agents, and workflows, helping teams operate private and sovereign AI with approvals, traceability, and consistent safeguards.
Runtime guardrails and cyber security inspectionAuditable governance and compliance as workflowControlled observability, logging, and tracing
Hardware foundation
NVIDIA DGX Spark, powered by the GB10 Superchip.
Galene.AI Elettra is built on the original DGX Spark Founders Edition hardware, using the NVIDIA GB10 Grace Blackwell superchip as the desk-side compute core of the appliance.
Exploded hardware viewNVIDIA DGX Spark
Built on NVIDIA
DGX Spark, powered by the NVIDIA GB10 Superchip.
The original DGX Spark Founders Edition brings NVIDIA Grace Blackwell compute into a compact desk-side form factor. Galene.AI Elettra uses that GB10 hardware core as the production foundation of a private and sovereign AI appliance.
Local memory
128 GB unified memory
Large memory footprint perfectly suited to run medium sized Mixture-of-Experts models with good performances
Parallel execution
Multi-model agentic flows
Keep multiple small and medium models active in parallel for more complex agentic AI use cases
Blackwell inference
NVFP4 quantization
Better inference efficiency and lower memory footprint while preserving model quality
Disclaimer
The current appliance target is 1 to 9 users, with 2 concurrent users as the expected full-speed operating point without noticeable slowdowns. Actual concurrent-user speed depends on the AI models in use and on the specific workloads being executed.
Future hardware option
DGX Station GB300, when more headroom is needed.
DGX Station GB300 is the future larger-format option for this appliance line when a desk-side setup needs more model headroom.
Future option
DGX Station GB300
A future larger-format step built around the NVIDIA GB300 superchip.
Hybrid scenario
Desk and department
DGX Spark can stay close to users on the desk while DGX Station GB300 can support broader departmental AI workloads.
Hint: DGX Station GB300Future option
Channel partner
Built in collaboration with TD SYNNEX.
Galene.AI Elettra on NVIDIA DGX Spark Founders Edition is built in collaboration with TD SYNNEX, who helps turn original NVIDIA hardware and the Galene.AI software layer into a single production-ready private and sovereign AI appliance.
TD SYNNEX supports the integration of the bundle as a deployable appliance while strengthening enterprise distribution and broader availability.
Private and sovereign AI, without the integration project.
If you want the original NVIDIA DGX Spark capability wrapped in a product experience built to bring private and sovereign AI into production, with cloud-like capabilities on your desk, Galene.AI Elettra is the right starting point.