The Key That Can’t Be Stolen

--

We’re ending the stolen-key category.

Press enter or click to view image in full size

Hack what?

That is the honest answer to the question the banks are asking this month. We’re not offering a clever answer. We’re providing an architectural one.

The banking sector is waking up to something it has known for a decade and declined to name: every wallet, every custody platform, every cold-storage rig on earth protects wealth with the same object. A secret stored as data. Data is infinitely copyable. That is not a flaw in the system — that is the system. The surprise was never that the attacks would come. The surprise is that we built a trillion-dollar asset class on an architecture where the secret must exist, somewhere, in a form a computer can read.

Stronger passwords did not solve this. Hardware wallets did not solve this. Multi-party computation did not solve this. Threshold signatures did not solve this. Every one of those approaches relocates the data. None of them eliminates it. The secret still exists. The attacker still has a target. The category still has a center.

—

What if the key never existed as data?

Not a better vault. Not a stronger algorithm. Not a more clever way to split the secret across more machines. My question is prior to all of that. I’m asking whether the thing being stolen must exist at all.

For forty years of digital cryptography, the answer was yes. It had to exist, because transistor-based systems offered no other way to prove identity, sign a transaction, or attest to a moment in time. Transistors store. Transistors switch. Transistors copy. The transistor is the medium of data — and an architecture built on transistors will always produce a key that is data, because the transistor was not designed to produce anything else.

The Poovey Switch was.

—

In April 2026, True Photonic filed the Photon-Locked Wallet architecture. It rests on three photonic primitives, each of which sits inside the SIGIL agent-trust infrastructure we filed earlier this year.

The first is the Photonic Physically Unclonable Function. It is a direct consequence of the Poovey Switch architecture — signing capability lives in the physical optical properties of the substrate itself, a fingerprint formed at the moment of fabrication, unique to that device, and impossible to extract because there is nothing to extract. The signing capability is the substrate. You cannot copy the substrate any more than you can copy a snowflake.

The second is Femtosecond Temporal Attestation. Every authorized action carries a time signature measured in femtoseconds — a speed domain that is 10,000X faster than a transistor. The windows in which replay attacks, forgery attempts, and timing games operate close before the attacker’s signal has finished leaving the room.

The third is the Photonic Quantum Random Number Generator. Randomness harvested from quantum optical behavior — not pseudorandom, not algorithmically seeded, not predictable by any classical or quantum adversary, because there is no seed. The randomness is native to the substrate.

Put those three together, and the consequence is architectural, not incremental. No seed phrase. No recovery phrase. No private key existing as data anywhere — not in software, not in firmware, not in an enclave, not in a vault, not in a backup, not in cold storage, not in a safe deposit box in Switzerland. Nowhere.

That is not a security improvement. That is a category ending.

The banks architecting the digital future do not need a better wallet. These banks need a custody architecture in which the question “was this key stolen?” cannot be structurally asked — because the object the question refers to does not exist in copyable form.

The Photon-Locked Wallet is designed to anchor the four primitives every financial institution is actually trying to protect.

Custody. Wealth under institutional control, signed by a substrate no adversary can clone, at a moment no adversary can falsify, with randomness no adversary can predict.

Settlement. Transactions authorized at the speed of light, gated in hardware rather than permitted in software — which means the window between authorization and execution is too short for the interception models every existing custody platform is designed around.

Identity. An institution’s signing authority tied to the physical device, not to a credential that can be phished, lifted, or exfiltrated. The credential is the thing. The thing cannot be copied.

Transaction authorization. Every action carrying a femtosecond-grade time signature, which means the compliance and audit questions that currently take weeks to resolve are resolved at the moment of authorization.

None of this is a patch on the existing architecture. It is a different architecture.

The tokenization architecture of the next monetary substrate is being finalized right now. Central banks are at the table. The commercial institutions that will clear, custody, and settle on that substrate are being chosen alongside it.

The window in which that decision can be made differently is measured in months, not years. This article is not an argument about that window. It is a notation that it is open.

Every wallet on earth today is a container for a secret.

Ours is not a container. There is nothing inside to steal.

Every bank currently architecting a digital-asset custody platform is building on a substrate that will be obsolete before the platform ships. The hacks will continue. The breaches will escalate. The board presentations explaining them will get harder.

The only custody architecture that ends the category is the one where the key was never data to begin with.

I was at this year’s Spring Meetings in DC. We’ve been getting the word out. Many now know the architecture exists. They know the IP has been filed. The question is which institutions will have heard and heeded — and which will be explaining, two years from now, why they didn’t.

The government once offered AT&T the backbone of what became the internet. The CEO did not see cause to take it.

Decisions.

Derek out.

Derek W. Bailey is Founder of True Photonic, Inc. and author of Keep Computing: How Light Solves Computing’s Impossible Problem (True Photonic, 2026). Contact: [email protected].

Disclosure: The author has a financial interest in photonic computing technologies. Statements regarding performance, capabilities, and market projections are based on preliminary research, testing, and simulations. Actual results may vary.

Forward-Looking Statements: This article contains forward-looking statements regarding future technology development and market conditions. These statements involve risks and uncertainties that may cause actual results to differ materially from those expressed. Readers should not place undue reliance on these statements beyond the date of publication.