Essay ยท August 2026

The Rails Arrived While Nobody Was Looking

Government credentials are landing at scale, some of it by legal mandate. The half of the transaction that accepts them is close to empty.

For about a decade, digital identity was a thing people had opinions about. Standards bodies argued, pilots launched and quietly ended, and every conference had a panel. The credential itself was always eighteen months away.

It is not eighteen months away. It is in fourteen US states and Puerto Rico, in the phone in your pocket, and in a European legal deadline that lands this December.

Take the US first. Driver's licenses and state IDs are live in Apple Wallet in Arizona, Arkansas, California, Colorado, Georgia, Hawaii, Illinois, Iowa, Maryland, Montana, New Mexico, North Dakota, Ohio, and West Virginia, with Puerto Rico alongside them. Several more have been announced. Separately, and more interestingly, Apple launched Digital ID built from a US passport, which sidesteps the state DMVs entirely and gives a credential to anyone holding a passport rather than only to residents of participating states. It went live at TSA checkpoints in more than 250 airports, and Apple has said the acceptance surface will widen to businesses, apps, and online.

Then Europe, which is the part most people outside it have not registered. Under eIDAS 2.0, all twenty-seven member states have a legal obligation to make a digital identity wallet available to their citizens by December 2026. Not a recommendation, an obligation. Some states will land late and some will land thin, and that matters less than it sounds like it should. The direction is fixed by law rather than by an adoption curve, and a few hundred million people are going to end up holding a government-issued credential on a phone.

So the issuing side is arriving, some of it by mandate. Here is the asymmetry that nobody seems to be pricing.

Almost all of the attention, the funding, and the standards energy has gone to issuance and to wallets. Who mints the credential, who holds it, what it looks like on screen. The other half of the transaction, the part where somebody actually checks the thing, is close to empty. Outside of airport security and a handful of age-verification pilots, there is very little that can accept one of these credentials and do something useful with the result.

The scarce thing has flipped.

For a decade the hard part was getting a trustworthy credential into someone's hand. That is being solved, at scale, by governments and platform vendors with resources nobody else has. What is scarce now is somewhere to present it. That is a strange position for an industry to find itself in, and it is the whole opportunity.

The good news is precise and worth stating exactly: you can build the accepting side today, and it requires permission from nobody.

We proved that on a Wednesday in July. The European reference wallet on an iPhone, holding a credential issued live by the EU's own service, presented through Safari's Digital Credentials API to a verifier we wrote, which decrypted it, checked the device binding, verified the issuer chain to the European root authorities, and created a session. No Apple approval. No agreement with any wallet vendor. No relationship with the issuer beyond holding their public trust anchors, which are published. The browser API is there, the standard is real, and an independent implementation interoperated with ours on the first serious attempt.

The honest complication is Apple, and it cuts the other way. Reading a credential out of Apple Wallet is not open in the same sense. Certificates gate whether you can request at all, and the request itself has to be signed with a Business Connect certificate, which means an application, a review, and a timeline that belongs to Apple. We have that submitted. It is not fast.

I do not think that changes the conclusion. Apple holds the credentials most Americans actually carry, so anyone serious about the US wants that rail and will wait for it. But the proven path went through a third-party wallet and needed no authorization, which means the protocol work, the decryption path, and the verifier can all be built and tested now, with Apple slotting in later as a second source rather than as a rewrite. An architecture that cannot start until a platform vendor approves you is a bad architecture regardless of how the approval goes.

One caveat about what acceptance actually means, because it is easy to hear "verify the credential" and think the job is done. A government credential answers who you are once, at high assurance, in an interaction designed to be attended and deliberate. Nobody presents a passport forty times a day. So verifying it is the front door, and everything the person does afterward needs credentials minted from that moment rather than repeated presentations of it. That downstream layer is a separate piece of infrastructure and a separate argument, which I have made elsewhere.

For the purposes of timing, the point is narrower. The credential problem is being solved by institutions with the scale to solve it. The acceptance problem is not, and it is wide open precisely as a few hundred million credentials are about to land in people's pockets.

The panels were not wrong about the destination. They were just wrong about it being far away.

All essays