DeltaBlue Sovereign (AI) Platform

Your AI. Your data.
Provably in the EU.

Run open-weight models, private document search and AI applications on isolated GPU capacity, operated by a Belgian, ISO/IEC 27001:2022 certified provider. Every request leaves an audit trail you can hand to your auditor. Your data never leaves EU jurisdiction.

Talk to us
  • ISO/IEC 27001:2022 certified
  • Operated from Belgium
  • EU jurisdiction only
  • No lock-in

Request record

Attested
Tenant
[tenant-id]
Model
Qwen/Qwen3-8B
Processed at
Edge site, BE
Jurisdiction
EU
Hardware
Measured boot, verified
Prompt stored
No
Output stored
No
Output fingerprint
Keyed hash only

A verifiable receipt for every request

Only request metadata and a keyed hash of the output enter the ledger. Prompts and generated content are never stored in it.

  1. 01Hash outputCreate a keyed fingerprint
  2. 02Sign receiptSeal metadata and fingerprint
  3. 03Verify proofRecompute from the original output

AI on sensitive data, without sending it abroad

Most AI services run on platforms that answer to non-EU law. For hospitals, governments, banks and anyone with confidential IP, that is a blocker. DeltaBlue gives you the same AI capabilities on infrastructure you can prove is European.

EU law, end to end

Operated by a Belgian company on EU infrastructure. Data, model weights, prompts and outputs stay under EU jurisdiction.

Proof, not promises

Every request is recorded with where it ran, on which attested hardware and with which model. Your DPO gets evidence, not a slide.

AI where the power is

Capacity runs on secure edge sites sized to grid connections that already exist, so you get GPUs without waiting for a new data-centre campus.

The platform

Everything you need to run AI in production

Isolated GPU inference

Dedicated or securely shared GPUs with hardware-level tenant isolation. No shared memory or model state between customers.

Private RAG workspaces

Let AI answer from your own documents. Ingestion, embeddings and vector search stay encrypted per tenant, with keys you control.

Verifiable sovereignty

A tamper-evident record per request: location, hardware attestation, model version. Prompts are never logged by default.

Secure edge sites

Run AI in our data centre, in an existing data room or on your premises. Measured boot, tamper detection and remote attestation on every node.

Open-weight models of your choice

Deploy the open models you trust, from compact assistants to large reasoning models, and switch when something better comes out.

Energy-aware by design

Batch jobs run when the grid is greenest. You get energy and carbon figures per workload for your CSRD reporting.

How it works

One control plane, every site

You call one API. Our control plane places each workload on the right site by latency, data residency and available power, with the same isolation and audit everywhere.

EU jurisdiction, operated by DeltaBlue

Your apps and developers

Self-service portal and AI-compatible API

Control plane

Placement by latency, residency and site power · per-request audit trail · customer-held keys · energy metering

Central platform

DeltaBlue data centreIsolated GPU tenantsConfidential RAG

Edge site

Your premisesIsolated GPU tenantsAttested, tamper-detected

Edge site

Existing data roomSized to the local gridAttested, tamper-detected

Decentralised by design

GPU capacity that doesn't wait for the grid

In Europe, power, not space, decides how fast AI capacity can grow. New large campuses wait in line for heavy grid connections. DeltaBlue spreads AI over smaller secure sites that fit the capacity already available, close to you and your data.

Capacity now, not in years

Each edge site is sized to an existing connection, so new capacity comes online site by site.

The same guarantees everywhere

Edge nodes carry the same isolation, attestation and audit trail as our central platform.

Data stays local

Process sensitive data on your own premises or in your region, managed by our team 24/7.

BYOGPU · Bring Your Own GPU

Give idle compute a new source of income

Customers, infrastructure owners and gaming enthusiasts can contribute spare CPU and GPU power to DeltaBlue. Earn revenue when eligible workloads run on the capacity you provide.

You bring the hardware and choose the capacity and hosting windows you can offer. DeltaBlue brings workload placement, usage metering and a shared platform for providers and customers.
Talk to us

Tell us about your hardware and where it runs.

The provider journey

Your hardware. A shared compute network.

  1. Your idle CPU & GPU capacity

    From a single machine to a fleet of servers

  2. Matched to eligible workloads

    Based on hardware, location and availability

  3. Metered usage. Provider revenue.

    Track contributed compute and earnings in the platform

Revenue follows actual paid usage. Hardware eligibility, availability and commercial terms are agreed during onboarding.

Customers

Make spare capacity in your existing environment available between your own workloads.

Infrastructure owners

Connect underused GPU servers or entire racks and turn spare capacity into a new revenue stream.

Gaming enthusiasts

Put compatible desktop hardware to work during idle periods. Choose hosting windows around the time you need your machine.
01

Register your hardware

Share your CPU, GPU, memory, connectivity and hosting location.
02

Set your availability

Choose the capacity and hosting windows you can reliably provide.
03

Run eligible workloads

Jobs are matched to your hardware, location and security profile.
04

Track usage & earnings

Review metered CPU/GPU time and the revenue it generates in your provider view.

The right workload on the right hardware

Provider location and security controls determine which jobs a machine can run. Community capacity serves suitable workloads; sensitive workloads stay on approved infrastructure that meets their residency, isolation and attestation requirements.

For developers

Switch your endpoint, keep your code

Our API works with the AI client libraries your team already uses. Point them at DeltaBlue and your application runs on sovereign infrastructure, with no rewrite.

Read the docs
from openai import OpenAI

client = OpenAI(
    base_url="[YOUR DELTABLUE ENDPOINT]",
    api_key="[YOUR API KEY]",
)

reply = client.chat.completions.create(
    model="Qwen/Qwen3-8B",
    messages=[{"role": "user", "content": "Summarise this contract"}],
)

Inside your platform

From individual requests to evidence you can share

See how sovereignty reporting could bring request receipts into the same workspace as your availability and resource-usage views. Choose a period, inspect the evidence and prepare your audit handover.

DeltaBlue / Platform reportsInteractive preview · sample data

Your workspace, in view

Explore the reporting workflow with illustrative records.

Trace a request to its processing site and inspect the metadata and output fingerprint recorded in the ledger.

Jurisdiction
European Union
Ledger contents
Metadata + keyed hash
Prompts & outputs
Outside the ledger
Request evidence: illustrative records for All environments, last 30 days
WorkloadEnvironmentProcessing siteLedger receiptDetails
Document assistantProductionEdge site, BEMetadata + output hash
Knowledge searchStagingCentral platform, EUMetadata + output hash
Contract reviewProductionCentral platform, EUMetadata + output hash

Document assistant · Request evidence

Processed at
Edge site, BE
Model
Qwen/Qwen3-8B
Output fingerprint
HMAC-SHA-256
Verification
Original output + authorised key

This preview contains no live customer data. The sample download is JSON; the platform’s availability report exports to PDF.

Choose your scope

Use environment and date-range filters to focus your review on the workloads and period that matter.

Inspect the evidence

Follow a request to its processing site and inspect the ledger metadata and output fingerprint, without exposing content.

Prepare your handover

Bring request evidence, availability history and resource usage together for your operations and compliance teams.

Built for organisations that can't compromise on data

Healthcare and life sciences

Clinical notes, research data and patient records, processed under GDPR.

Public sector

Citizen services and internal assistants on infrastructure the government can audit.

Financial services

AI that supports your DORA and outsourcing requirements.

Legal and professional services

Contract review and knowledge search without client data leaving the EU.

Industry and manufacturing

Engineering documents and IP stay on site, on an edge node in your own plant.

AI-native SaaS

EU-hosted GPU capacity your enterprise customers will sign off on.

Plans

Start small, scale on your terms

Shared isolated inference

Pay per use on securely shared GPUs. Ideal to start.

Dedicated GPU nodes

Managed GPU nodes reserved for you, centrally or on an edge site.

Private RAG workspace

Secure search and Q&A over your own documents.

Onboarding and integration

Our engineers set up your models, data and integrations with you.

Evidence for your compliance work

Platform controls are mapped to the frameworks you report on, and we deliver the evidence your auditors ask for.
  • ISO/IEC 27001:2022
  • GDPR
  • EU AI Act
  • NIS2
  • DORA

Put your first private model to work

Tell us your use case. Our engineers set up a pilot on your data, in your chosen location, with the audit trail switched on from day one.

Talk to us