OPAQUE, The Confidential AI Company

AI is about to touch the most human things we have — our health records, our finances, our private conversations, the decisions that shape our lives. The promise is extraordinary. But it arrives with a quiet bargain almost no one agreed to: to use AI, you hand over your data and hope it's handled well.

We started OPAQUE to break that bargain. The price of AI should not be your data.

Every hospital, bank, and government putting AI near something that matters asks the same question first: Can I trust it? Today the honest answer is "you'll have to take our word for it." Faith is not a security model. Trust shouldn't be a promise, a policy document, or a vendor's reputation — it should be something you can verify, enforced in hardware and backed by proof, every single time.

And every honest builder knows the deeper reason this matters: these systems leak. A chatbot spills context onto servers no one controls. Agents are worse — give one real reach and even a one-percent chance of leaking, run a thousand of them, and it stops being might. It will. That isn't a flaw to apologize for; it's the nature of the technology. The answer isn't to trust harder. It's to make trust unnecessary — to prove the data stayed protected no matter what the system did.

In June 2026, Apple proved that's possible at a scale no one had attempted: verifiable, private AI across a billion iPhones — fast enough and cheap enough for consumers, running even on hardware Apple doesn't own. If it can be done on the phone in your pocket, your bank can do it for your trades and your hospital can do it for your chart. The excuse that verifiable privacy is too slow or too expensive is gone. Apple drew the line. Our job is to put it within reach of everyone else.

That conviction is the whole company. When trust can be *proven* instead of promised, the doors bolted shut today swing open: regulated industries, the most sensitive data, agents that take real action in the real world. Verifiable trust is how AI finally connects to the systems that run the world — safely, and on terms you can check yourself.

We build the layer that makes that possible.

What's really at stake

This is bigger than any one product, any one deal — bigger than OPAQUE.

Data sovereignty — the ability of a person, a hospital, a company, or a nation to run AI on its own data, on its own terms, without surrendering control — is becoming a fundamental right. Not in a woo-woo sense. In a load-bearing one.

Because if sovereignty fails, AI doesn't just get riskier — it consolidates. The reasoning engine behind every diagnosis, every contract, every customer conversation collapses into one or two companies that power the cognitive infrastructure of the entire world — every industry, every government, every institution renting its ability to think from a handful of owners who answer to no one. That isn't a market. It's dependence at a scale civilization has never run before.

We don't believe that future is inevitable — and preventing it is the work. When data stays sovereign, the global AI economy stays open, competitive, and plural: thousands of institutions building on their own data instead of a few platforms building on everyone's. A healthy economy, stable institutions, a free society — each now rests on getting this one thing right. We think it's the most important problem in technology this decade. It's why we're here.