The Ownership Spectrum

You rent your AI.
Here's what owning it looks like.

Most businesses pay per token for intelligence they don't control. There's a better way — and it doesn't require a data center.

Three levels

From renting to sovereign

Most businesses end up at Level 2 and never need Level 3. Read across; pick the one that fits your obligations.

Level 1

Renting

What most businesses do today.

You send prompts to OpenAI, Anthropic, or Google. They process your data on their servers. They set the price. They set the terms. They can change both whenever they want.

You don't own the model.
You don't control your data.
You can't switch without rebuilding everything.

Examples: ChatGPT API, Claude API, Gemini API
Cost: $0.01–0.06 per 1K tokens (adds up fast at scale)
Risk: Vendor lock-in, price changes, model deprecation, data exposure

→ This is where most businesses should start.

Level 2

Owned

The practical middle ground. Where most businesses should be.

You download an open-weight model — the actual files. You run it on any cloud provider you choose: GCS, AWS, Azure, DigitalOcean. If your provider raises prices, you move to another one over a weekend. The model is yours. The weights are yours. The fine-tune is yours.

Your data stays in your environment.
No vendor can deprecate your model, change your pricing, or train on your prompts.
Cloud is commodity infrastructure — the intelligence is yours.

Examples: SimpleDirect flash-1 on GCS, Llama on AWS, Mistral on Azure
Cost: Cloud GPU rental ($1–5/hr) — predictable, no per-token surprise
Risk: Minimal — model is portable, provider is interchangeable

Level 3

Sovereign

For organizations where data residency and jurisdiction matter.

Everything in Owned, plus:

Runs in a specific country's data centers (Canadian for ISED, PIPEDA, defence).
Can be deployed air-gapped with zero external network dependencies.
Meets regulatory and classification requirements (Protected B, ITAR, etc.).
Full chain of custody: you know where your data is, physically.

Examples: SimpleDirect flash-1-pro on Telus (Rimouski, QC), on-prem colo, air-gapped rack
Cost: Sovereign cloud ($3–8/hr) or colocation ($5–15K/mo)
Risk: Near-zero — you control every layer of the stack

How to choose

Three questions. One answer.

Walk through the questions. The first "no" tells you where to land.

Question 1
Do you handle regulated data?
Client files, patient records, financial data, classified information, anything under NDA.
No
Level 2 (Owned) is probably enough.
Yes
Keep going.
Question 2
Does your regulator care where the data physically lives?
PIPEDA, PHIPA, HIPAA, ITAR, Protected B, Five Eyes restrictions.
No
Level 2 (Owned) on a reputable cloud provider covers you.
Yes
Keep going.
Question 3
Could your organization face audit, breach liability, or classification requirements?
Government contracts, defense work, healthcare systems, legal privilege.
No
Level 2 (Owned) with good security practices.
Yes
Level 3 (Sovereign). You need jurisdiction control and possibly air-gapped deployment.

Most businesses land on Level 2 and never need Level 3. That's fine. Owned is a massive upgrade from renting.

Side by side

What SimpleDirect offers at each level

Same files. Different deployment posture. Pick the one that fits your obligations.

	Renting (Level 1)	Owned (Level 2)	Sovereign (Level 3)
Your role	Consumer	Operator	Controller
Who has your data	The API provider	You (on cloud you choose)	You (on hardware you choose, in jurisdiction you choose)
Model portability	Zero — locked to provider	Full — GGUF runs anywhere	Full — same files, your rack
Price control	None — per-token, they set it	You choose cloud provider	You own the hardware
Can you be cut off?	Yes — account suspension, API deprecation	No — you have the files	No — you have the files AND the hardware
SimpleDirect model	N/A	flash-1-mini (free), flash-1 ($99), flash-1-pro ($499)	flash-1-pro ($499) + deployment support ($10–25K)
Where it runs	Their servers	GCS, AWS, Azure, DigitalOcean, any cloud	Telus (QC), Canadian colo, on-prem, air-gapped

Renting (Level 1)

Your role: Consumer
Who has your data: The API provider
Model portability: Zero — locked to provider
Price control: None — per-token, they set it
Can you be cut off?: Yes — account suspension, API deprecation
SimpleDirect model: N/A
Where it runs: Their servers

Owned (Level 2)

Your role: Operator
Who has your data: You (on cloud you choose)
Model portability: Full — GGUF runs anywhere
Price control: You choose cloud provider
Can you be cut off?: No — you have the files
SimpleDirect model: flash-1-mini (free), flash-1 ($99), flash-1-pro ($499)
Where it runs: GCS, AWS, Azure, DigitalOcean, any cloud

Sovereign (Level 3)

Your role: Controller
Who has your data: You (on hardware you choose, in jurisdiction you choose)
Model portability: Full — same files, your rack
Price control: You own the hardware
Can you be cut off?: No — you have the files AND the hardware
SimpleDirect model: flash-1-pro ($499) + deployment support ($10–25K)
Where it runs: Telus (QC), Canadian colo, on-prem, air-gapped

The transition path

You don't have to jump. You can walk.

Most organizations move through the spectrum gradually.

1
Start with flash-1-mini on your laptop.
Test it. See if the quality works for your use case.
Total cost: $0. Total time: 5 minutes.
2
Move to flash-1 on a cloud GPU.
Run it for your team. Compare quality to your current API.
Total cost: ~$200/month on GCS or AWS.
3
Deploy flash-1-pro on Canadian infrastructure.
Telus, GCS Montreal, or your own rack. We help with deployment. Fixed fee, 2–3 days.
Only if you need sovereignty. Each step is reversible. Same model format throughout.

There is no lock-in at any level — including with us.

Pick your next step

For business

Ready to own your AI?

Start with the free model. No account, no API key, no commitment.

Get flash-1-mini — Free

For enterprise & government

Need sovereign deployment?

Tell us about your environment. We respond same business day. Bilingual EN / FR.

Talk to us

You rent your AI.Here's what owning it looks like.