D-Central Technologies — decentralized computing powering your digital sovereignty: Bitcoin, AI, Mesh, and the rest of the stack you can actually own.

Own your AI before someone switches it off

On June 12, 2026, a US export-control directive forced Anthropic to disable its Claude Fable 5 and Mythos 5 models for every foreign national worldwide. People who had built their work on those models lost access overnight, by directive, with no appeal. It was the clearest proof yet of a simple truth we have argued since 2016: if you do not own the compute, you do not own the capability. A login is a permission, and permissions get revoked.

We will not sell you a guarantee that nothing will ever be taken away. We will help you build a stack that does not depend on permission in the first place — capable AI on hardware you own, on power you control, that no one can switch off by decree. That is what AI Sovereignty Consulting is.

Book a Sovereignty Briefing →

The problem we solve

• Kill-switch & lock-in risk. Vendor terms change, models get deprecated, access gets geo-blocked or revoked. If a directive in another country can zero your week, you are renting, not owning.
• Privacy & Quebec Law 25. Every cloud AI prompt containing personal information is a cross-border data transfer, usually to US-owned servers reachable under the US CLOUD Act. On-premise AI removes that exposure by construction. (See our guide to Law 25 and on-premise LLMs.)
• The competence gap. The open-source tooling is genuinely good now — but getting from “I read a blog post” to “this runs reliably on the right hardware, sized and cooled properly” is full of VRAM math, quantization trade-offs, drivers, and power-and-heat realities we deal with every day.

Who it’s for

• The sovereign individual. The Bitcoiner, homelabber, or privacy-minded professional who already runs their own node and wants to run their own model the same way.
• Business & regulated firms. Law, accounting, medical, notary, agencies — anyone with confidential data and Law 25 exposure who needs capable AI that never leaves the building.
• Government, enterprise, co-ops & mining operations. Organizations that need data residency and resilience, or miners with power and space who want to turn idle capacity into sovereign compute.

One stack, five layers

AI is the layer that just made the news, but it is only one of five. Real digital sovereignty means owning enough of the stack that no single actor has an off-switch over your operation:

• AI — local, open-weight models on hardware you own.
• Bitcoin — money that settles on rules you can verify, with keys you hold.
• Mesh — radio and mesh networking (Meshtastic and beyond) that keeps you connected when the ISP layer is down or filtered.
• Nostr — keypair identity and publishing that no platform can revoke.
• Solar — energy you generate, the last off-switch moved into your own hands.

We consult across the whole stack. Most engagements start with AI because that is the urgent one — but if you want the mesh layer designed, a node stood up, or power planned around a Hashcenter, that is the same conversation. See the sovereign stack for how the pieces interlock.

How we work with you

Every engagement is hand-scoped. We don’t bolt you onto a template, and we don’t quote a number before we understand the job. Pricing is by quote — you’ll get a clear scope and a straight figure before anything starts.

1 · Sovereignty Briefing

A focused strategy session: your use case, your threat model, your budget. You leave with a plain-English, written recommendation — is local AI right for you, which model class fits (from an 8 GB box to a 70B-class workstation), and what hardware that implies. Yours to keep whether or not you build with us. This is the map, not the journey.

2 · Pleb Sovereign Build

For the individual. We size and source the hardware to your budget, install the stack (Ollama or LM Studio plus a private Open WebUI), load and tune an open model chosen for your job, and do the power-and-heat reality check that keeps it livable. Optional: wire in Nostr identity and a Lightning-ready node.

3 · Business On-Prem AI

For the office. Everything above, scaled to a team: a needs assessment, private RAG over your own documents, air-gapped or LAN-only deployment so nothing leaves the building, a staff-facing UI, and a short operator runbook so you aren’t dependent on us to turn it on. The cleanest answer to Law 25 we know of. Ongoing support available.

4 · Hashcenter Sovereign Compute

For enterprise, government, co-ops, and miners. Multi-GPU or high-VRAM design, optional colocation in a D-Central Quebec facility on clean hydro power, power and cooling engineering (including off-grid and solar), and an honest assessment of which existing mining hardware can — and can’t — become useful AI compute. Mesh and Nostr fallback for control-plane independence where it matters.

The process

1. Briefing — we understand the job and your threat model.
2. Model & hardware spec — the right model and the right silicon for your actual workload, sized honestly.
3. Sourcing & build — hand-built to order in Laval. Because we keep inventory lean and build each unit properly, lead time is an estimate, not a guarantee — we’ll confirm your window when the scope is set. That’s the craftsmanship, not a delay.
4. Install & handover — it arrives running, with a runbook and the keys in your hands.
5. Support (optional) — quarterly check-ins, model refreshes as better open weights ship, capacity reviews.

Why D-Central

We did not write Ollama, train Qwen, or invent llama.cpp — and we say so plainly. That work belongs to the open-source community and the open-model teams, and we stand on their shoulders. What we bring is judgment and integration: matching the right pieces to your real workload, on hardware we know how to size, power, cool, and keep running. We are a real Quebec shop with hydro, facilities, repair benches, and years on the hardware — not a faceless storefront. Not Amazon. A craftsman who picks up the phone.

Honest limits

Local and open models trade frontier-peak capability for control and continuity. On the very hardest tasks, the best hosted models are still ahead — by a margin measured in months, and closing. For the overwhelming majority of real work, a well-chosen local model is already more than good enough, and it is yours. We will tell you honestly when a cloud tool is the better call for a specific job, and we won’t oversell you a rig you don’t need. Read the straight comparison in local AI vs cloud AI.

Frequently asked questions

Is local AI as good as ChatGPT or Claude?

Not at the absolute frontier today — and we won’t pretend otherwise. For drafting, summarizing, classification, coding assistance, and private Q&A over your own documents, a good local model is already excellent, and it can’t be revoked. The gap on the hardest tasks is real but narrowing.

What hardware do I actually need?

It scales with the model: roughly an 8 GB GPU for a 7–8B model, ~24 GB (a used RTX 3090 is the value pick) for 30–32B, and ~48 GB — or Apple Silicon with 128 GB unified memory — for a 70B. We size it to your job, never the other way around. Prefer to buy it ready to run? See the Sovereign AI boxes.

Can you do it fully air-gapped?

Yes. For the most sensitive work we deploy a machine with no internet connection at all — nothing in, nothing out — with your files indexed locally so data physically cannot leave.

Does this satisfy Quebec’s Law 25?

On-premise AI removes the cross-border transfer and third-party-access problems by design, which is the cleanest posture available. Compliance is something your whole organization earns, not a single tool — but this makes it far more achievable. (Orientation, not legal advice.)

Can my old miner become an AI node?

Sometimes yes, often no — and we’ll give you the honest engineering answer rather than the one that sells more. Where a purpose-built GPU node is the right call, we’ll tell you, and your existing power and cooling may still be the head start.

Do you only do AI?

No. AI is the urgent layer right now, but we consult across the whole sovereign stack — Bitcoin, mesh (Meshtastic and in-house builds), Nostr, and the power underneath it. If you want to decentralize more than your compute, that’s the same conversation.

Book a Sovereignty Briefing

Tell us what you’re trying to run — or what you’re trying to stop depending on. Mention whether you’re an individual, a business, a government or enterprise team, or a mining operation, and roughly where you’re located (it matters for Law 25 and for colocation). We reply on business days and take it from there. No newsletter, no list, no spam — this message reaches our team and nothing else.

Name *

Email *

Order number (optional)

Message *

Leave this field empty

Send message

Prefer the phone? Call 1-855-753-9997 or email support@d-central.tech — Laval, Quebec, Monday to Friday.

New to this? Start with Sovereign AI in Canada for the why, or the local-LLM playbook if you’d rather build it yourself first.