Organizations Can Finally Stop Choosing Between AI and Data Privacy
I've recently been involved in many customer conversations, and a problem keeps coming up with enterprise AI teams, especially with customers in regulated industries. It goes something like this: “We have highly sensitive or proprietary data. Our teams have identified many AI use cases that could transform how we operate and do our business. But the moment that data leaves our control even briefly, even just to be processed, we’re in trouble.”
Until now, the honest answer to that problem has been: you either take the risk or sit on the sidelines. But in those regulated industries, no one will take the risk. Data must stay on premises, and if the data can’t go to the AI model, the AI model needs to come to the data.
Now let me share another conversation that goes something like this: We can’t share our frontier models for on-prem deployments. This is our business. And no, we trust no one—be it the financial industry giants or the trillion-dollar enterprises.
That changes today.
TELUS and Fortanix are joining forces to enable organizations run AI on their most sensitive data without ever actually exposing it. Built on NVIDIA’s Confidential Computing infrastructure and deployed through TELUS’ Sovereign AI Factory, Fortanix Confidential AI uses confidential computing technology to ensure the end-to-end solution runs in a trusted environment, which is cryptographically verified before any sensitive data is ever decrypted. If the hardware check fails? The data stays locked. Full stop. “No Attestation No Key Release.”
What makes this different from “we’re secure, trust us.”
This is the part that matters if you’ve been in the room when compliance and legal teams push back on AI adoption. The usual conversation involves a vendor saying, “We take security very seriously” and pointing to a terms-of-service document. That’s not proof, that’s a contract.
Fortanix Confidential AI generates a cryptographic proof. Before a cryptographic key is released, NVIDIA’s attestation service verifies that the trusted execution environment is genuine and untampered with. That means the evidence isn’t just on paper, it’s embedded in the math. You can audit it. You can point to it in a regulatory conversation. That’s a fundamentally different posture.
Why sovereign infrastructure matters here
This isn't just a technical story; it's also a jurisdiction story, and for organizations operating under strict data residency requirements, that jurisdiction piece is everything.
Data privacy laws, whether regional, national, or sector-specific, aren't abstract compliance checkboxes. They're the reason AI projects stall before they ever get off the ground. Legal says the data can't leave the territory, and your AI project ends up stuck.
The TELUS Sovereign AI Factory resolves that standoff. The compute is physically on Canadian soil, operated by a carrier headquartered there, which means organizations aren't threading the needle between AI adoption and cross-border data transfer rules. And when you layer Fortanix's cryptographic attestation on top of that, you're not just claiming the data stayed put, you can prove it did.
That combination of sovereign location, hardware-enforced encryption, and auditable attestation is what makes this a viable solution for the organizations. Together, they form a verifiable chain of trust that covers the entire AI lifecycle: training, fine-tuning, and inference.
Who this is actually for
If you’re running a hospital network trying to use patient records to improve patient care or diagnostic models, this is for you. If you’re a financial institution that wants to fine-tune a fraud detection model on real transaction data without creating a compliance nightmare, this is for you. If you’re a government agency evaluating AI and your first question is “where does the data actually go?” this is for you.
The regulated industry AI problem has always been less about capability and more about trust. The models exist. The compute exists. What was missing was a credible answer to “how do I know it’s safe?” This is the answer.
The bottom line
Fortanix Confidential AI delivers end-to-end protection for proprietary frontier models, sensitive data, and inference across the full AI lifecycle. Enterprises can now run advanced AI on their data, while model providers keep their critical IP fully protected.
The TELUS-Fortanix partnership makes this available on sovereign infrastructure at scale. For organizations waiting to move forward with AI, this is it.
