Legal access
CLOUD Act and FISA 702: US authorities can reach data in US clouds, even if the server sits in Frankfurt.
BORDONARO AI · SOVEREIGN AI FROM THE ENTERPRISE LAB
Pull the plug.
Our AI keeps answering.
Generative AI on hardware you own: Azure Local, Foundry Local, NVIDIA GPUs, fully disconnected from the grid if you wish. No subscription that gets pricier every month. No foreign jurisdiction reading along. No model that vanishes overnight. Proven first in our own lab with 2,516 cores, then productive in your basement.
8x Microsoft MVP · Enterprise lab with 2,516 cores · Part of the Sovereign Cloud Alliance
◆ 8x Microsoft MVP◆ Azure Local◆ Foundry Local◆ NVIDIA H100 / L40S◆ Enterprise lab · 2,516 cores◆ Sovereign Cloud Alliance◆ VS-NfD preparable◆ EU AI Act · BSI C5◆ 8x Microsoft MVP◆ Azure Local◆ Foundry Local◆ NVIDIA H100 / L40S◆ Enterprise lab · 2,516 cores◆ Sovereign Cloud Alliance◆ VS-NfD preparable◆ EU AI Act · BSI C5
Microsoft MVP in a row, re-awarded every year
800+
professionals trained as a Microsoft Certified Trainer
2,516
CPU cores in our own enterprise lab (availability class 4)
30
years in IT, from the datacenter to the boardroom (almost)
All figures documented on dino.bordonaro.de, MVP status publicly verifiable at Microsoft.
„I love working with Dino. He's technical, professional, positive, passionate and curious. He has become one of the world's experts on Azure and Azure Stack that I go to when looking for recommendations on how we can make the product better.“
David ArmourPrincipal Product Manager Lead, MicrosoftAn independent Microsoft product engineering voice on technical depth. Not an AI project reference, and not presented as one.
WHY NOW
Your cloud AI is only as sovereign as a foreign law allows.
The US CLOUD Act compels providers to hand over data, regardless of where it is stored. Neither the EU Data Boundary nor privacy frameworks rule that out. Your biggest lever is your own hardware, which you can take off the grid.
Data leaves
Prompts, documents, customer data leave your building on every cloud call. Your contract no longer decides alone who reads them.
Loss of control
Model gone, price up, feature cut: with rented AI the vendor decides. With your own hardware, you do.
Data residency is not sovereignty.
A datacenter in Frankfurt changes nothing about foreign access rights, license dependencies and operational reality. AI becomes sovereign through architecture, operations and responsibility: owning the hardware, being able to disconnect and a team that masters the system.
THE SOVEREIGNTY SPECTRUM
"AI from your own basement" isn't a metaphor. It's an architecture decision.
The more regulated your organization, the further right your AI belongs. We build the sovereign end: with our own hardware and our own operations.
01
Public cloud
Convenient, but a foreign legal space. Data can leave, access is not ruled out.
02
Private cloud
More isolated, but still at the hyperscaler and still under foreign law.
03
On-prem
In your own data center. Full data control, no foreign cloud.
BORDONARO AI
04
Air-gapped
Disconnectable from the grid, preparable for requirements up to VS-NfD. This is where BORDONARO AI stands.
The difference is physical.
Pure AI platform (SaaS)
- Software you rent
- Runs in someone else's cloud
- no hardware of its own
- sovereignty promised by contract
- air-gap and defense not addressed
BORDONARO AI
- Infrastructure and products you own
- On-prem by default, disconnected operations as an option
- Azure Local plus NVIDIA edge GPUs, as an appliance
- Sovereignty is physical: in your building, disconnectable
- Defense-ready · certified lab · 8x MVP
A platform is good. Ownership is better.
HOW WE WORK
Four steps from use case to sovereign AI.
Discovery workshop
Your tasks, your data, your risks, taken honestly. A prioritized backlog at the end.
Pilot in the lab
The best use case is built and tested in the enterprise lab before you sign.
On-site rollout
Appliance in your building, assistant tailored to your processes, users onboarded.
Managed operations
Operations, updates, model care. You use the AI, we keep it in the air.
WHAT YOU CAN AUTOMATE
Most work doesn't happen inside your systems. It happens between them.
That's exactly what an assistant takes off your plate. The assistant drafts, the human approves. Click a task to see what's left and what it saves per year.
Sales
Create proposals
Before (manual)
45 min
After (with approval)
8 min
saved per task−37 min
Saved per year≈ 321 hrs
RUN THE NUMBERS · YOUR NUMBERS
What a single assistant gives back per year.
Every assumption is visible and editable. No marketing, just your numbers.
867
hours / year
21.7
work weeks
52,000
€ of working time / year
And what does it cost? The business case.
5.5 months
pays back after
24 %
ROI in year 1
10,120 €
Net savings year 1
Based on your assumptions. Defaults are deliberately conservative (60-90 % automation is realistic in practice).
Pressure-test this business case with usAN AGENCY · WITH REAL PRODUCTS
From use case to the GPU in the rack. One partner.
Product · RAG
Sovereign Assistant
On-prem AI assistant on Foundry Local. Your own knowledge base, roles and rights, air-gappable. The assistant drafts, the human approves.
For: all regulated organizations
Product · Voice
VOXA
Local transcription with speaker separation. Privacy at the root: nothing leaves the building. No US-cloud upload for sensitive conversations.
For: healthcare · law · public sector
Product · Hardware
Sovereign AI Appliance
Pre-configured Azure Local node plus NVIDIA GPU (L40S up to H100/H200), validated in the lab. The physical "AI in your basement".
For: ownership and air-gap
Product · Agents
Sovereign Agents
Workflow automation between your systems, orchestrated on-prem. Connectors within a clearly defined scope.
For: process automation
Consulting that builds instead of producing slides.
01
Assessment and architecture
Readiness taken honestly, target picture, reference architecture incl. GPU sizing. No slideware.
02
Implementation and rollout
Platform build, knowledge base, GPU deployment. Flown in the enterprise lab first, only then in your production.
03
Managed Sovereign Operations
Operations, monitoring, updates, model care. An escalation model, not finger-pointing. 24/7 optional.
04
Data and RAG engineering
Knowledge base build, retrieval, data quality: the basis for answers you can trust.
05
Compliance and governance
EU AI Act, BSI C5/AIC4, defense readiness, GDPR: as architecture and evidence, not as a PDF.
WHERE YOUR AI LIVES
It doesn't have to be your basement. Ours is in Mutterstadt.
Sovereignty means choosing where your AI runs. Three models, one responsible partner, the same architecture standards.
On-prem
In your own house
The Sovereign AI Appliance in your datacenter or server room: maximum control, disconnected from the grid if you wish. The royal road for highly regulated organizations.
Shared and dedicated
In our datacenter
Your AI runs in Datacenter Rhein-Neckar II by PFALZKOM in Mutterstadt, a TÜV/tekit-audited availability class 4 facility: as a shared AI environment for a fast start without own hardware, or as individual housing and hosting with dedicated systems. German jurisdiction, short distances, our team knows every rack.
Hybrid
Both, cleanly connected
Sensitive workloads stay with you, scaling inference runs with us, governed through one shared control plane. Switching between the models remains open at any time.
Same team, same standards, same contact person, no matter where the GPUs live. The packages below apply to all three deployment homes.
THREE PROOFS INSTEAD OF PROMISES
What has already worked under responsibility.
Three real stories from our work, honestly labeled: the first two are infrastructure and own-operations proofs, not AI customer projects.
Operations under responsibility
200+ dialysis clinics, 7,000 employees, 24/7
For KfH, the team around Dino Bordonaro rolled out a fully automated Azure Stack Hub platform across more than 200 dialysis clinics. Medical IT that treatments depend on, around the clock. Whoever has operated systems like that understands what an AI system must withstand before it is given responsibility.
200+ clinics · 7,000 users · fully automated rollout
Infrastructure reference from the company history, not an AI project.
Our own operations
Our AI operating system runs in our own company, every day
Proposals, receipts, knowledge and processes of BORDONARO IT run on a self-hosted, sovereign AI system. We sell nothing we do not operate ourselves: the Sovereign Assistant and VOXA are daily reality in our own house, not a demo.
Self-hosted · sovereign · productive daily
Own-operations proof of BORDONARO IT.
Scaling under pressure
The stress test: tens of thousands of workplaces into the cloud
When entire workforces could no longer come to the office overnight during COVID, the network around Dino Bordonaro helped move workplaces and applications for tens of thousands of users to Microsoft Azure across several customer environments. Crisis-proven migration experience our rollout methodology builds on.
Tens of thousands of users · several environments · under time pressure
Cloud migration reference from the company history.
WHERE WE COME FROM
Our AI story didn't start with ChatGPT. It started with systems that were not allowed to fail.
BORDONARO AI builds on almost 30 years of infrastructure, operations and Microsoft technology: medical IT for more than 200 dialysis clinics, Azure hybrid since the technical preview, an enterprise lab in a TÜV-audited datacenter and an AI operating system our own company uses productively every day. Whoever wants to give AI responsibility should have operated systems that carried responsibility first.
„Dino is an exceptionally talented blend of technology and business. He is passionate about doing the right thing, always ahead of new technologies, and has an innate gift for understanding how technology can help customers solve their problems.“
Vijay TewariDirector of Product Management, Google Cloud, formerly Microsoft Product Lead Azure Stack„Passionate, committed, opinionated and execution-strong. That characterizes Dino best. Whoever has Dino as a partner at their side can sleep well at night.“
Michael HollmannIT executive, 25+ years of IT leadershipSelected organizations from almost 30 years of IT work
- KfH
- BMW
- Mercedes-Benz
- Deutsche Bahn
- Deutsche Telekom
- Bayer
- ZEISS
- Schaeffler
- Wacker Chemie
IT, project and training engagements of Dino Bordonaro and BORDONARO IT. Deliberately not presented as an AI customer list. Defense projects exist under NDA.
BORDONARO AI ACADEMY
From the boardroom to the rack.
Most AI trainings end with a certificate. Ours end with something that works on Monday: 20 role-based tracks in four learning paths, from the board to the platform team, with agendas, fixed prices and artifacts.
Go to the AI Academy Start the academy diagnosis
18 tracks · 4 enterprise programs · certificate and training records included
THE LAB IS THE PROOF
Every architecture flies in the lab first. Only then does it touch your production.
The BORDONARO IT enterprise lab lives in Datacenter Rhein-Neckar II (PFALZKOM), a TÜV/tekit-audited availability class 4 facility. Architectures are built, deliberately broken and repaired there before they touch a customer system. Documented reference values, no live telemetry.
Rollouts run on our own Disconnected Operating Framework (DOF), proven in defense, critical infrastructure and healthcare environments.
2.516
CPU cores
24 TB
RAM
3.400 TB
storage
200 Gb/s
network fabric
8x Microsoft MVP in a row
Re-awarded every year since 2019, currently in Microsoft Azure and Cloud and Datacenter Management. A personal award of Dino Bordonaro, publicly verifiable.
Verify ↗Microsoft Certified Trainer and 19 certifications
Annually renewed trainer status, plus 19 Microsoft certifications from fundamentals to expert level, individually linked on dino.bordonaro.de.
Verify ↗TOP CONSULTANT 2020 · TOP 100 Innovator 2019 and 2020
Awards of BORDONARO IT GmbH & Co. KG from its founding era: 3rd place TOP CONSULTANT (size class A, based on reference customer surveys) and twice TOP 100 for innovation in the German Mittelstand.
Historical company awards (2019/2020), correctly attributed to BORDONARO IT at the time.
Verify ↗ENGAGEMENT MODELS
Transparent entry points instead of proposal poker.
Four project workshops to get started, three packages on the way to owning your AI. All prices net plus VAT, public, no asking required.
Project workshops
DISCOVERY
Use-case discovery
- Duration:
- 1 day
- Outcome:
- Prioritized use-case backlog and roadmap
from €4,900
ARCH
Architecture sprint
- Duration:
- 2-3 days
- Outcome:
- Reference architecture, GPU sizing, PoC scope
from €9,900
VSNFD
Air-gap / defense check
- Duration:
- 1-2 days
- Outcome:
- Readiness report and action plan
from €3,900
ROADMAP
Sovereign AI bootcamp
- Duration:
- 5 days
- Outcome:
- Strategy, architecture and delivery plan
from €18,500
Packages
Sovereign Pilot
from €24,000
one-off
- Discovery workshop (1 day)
- PoC in the certified lab
- Sovereign Assistant, set up on-prem
Own it
from €35,000
+ ops from €1,490/month
- Sovereign appliance on-site
- Assistant plus your own knowledge base
- 6 months managed basic included
Air-Gap · Defense
custom
quote by scope
- Defense readiness check
- appliance operable disconnected
- Enterprise operations 24/7
Day rate anchor €2,100 · Workshops €3,900-18,500 · Academy tracks from €3,950 · Managed from €1,490/month. Every package available at your site or as housing and hosting in DCRN II Mutterstadt. All prices net, market-based recommendations.
SOVEREIGN-AI CHECK
In 90 seconds: how sovereign and ready are you really?
Eight questions. Honest answers. A result with a clear recommendation.
THE NEXT STEP
Bring your AI back into your own house.
A 30-minute call where we tell you honestly whether and how sovereign AI pays off for you. No slideware.
I'll get back to you personally as soon as possible.