We mapped more than 500 identity companies.
The pattern across all of them is the same, and it is structural. Each one solves identity for exactly one entity type — your login, your KYC check, your wallet, your agent — inside a silo none of them can leave. Because the silo is the business model. Their revenue requires the credential to stay inside their walls. The moment it ports, the moat is gone.
So identity fragments three ways at once:
Humans get re-verified endlessly. The same passport, re-checked on every platform, stored in every database, exposed in every breach.
AI agents have no shared identity primitive at all. As agents begin to transact in 2026 — booking flights, renewing subscriptions, negotiating refunds, paying bills — there is no standard way to prove who issued an agent, what it is cleared to do, or which human (or organization) is on the other side of that delegation.
Machines face the same gap. Industrial machine-to-machine identity exists in isolated enterprise stacks; nothing portable, nothing standards-native.
This is one problem wearing three masks. And the fix is not a better silo. It is a shared primitive that any of the existing solutions can build on top of.
The convergence thesis
Identity, like networking and TLS before it, will converge on an open layer underneath the products. The question is who builds it.
A credible candidate has to do three things at once. It has to be one open credential primitive — a W3C DID plus a Verifiable Credential, with selective disclosure, signed by issuers a relying party can independently verify. It has to be issuable once and reusable everywhere, across humans and agents and machines, without re-verification per platform. And it has to be anchored in a layer that survives the failure of any single issuer or verifier.
That is the protocol-shaped problem. It is not solved by a SaaS company that owns the credential after issuing it.
Why this hasn't happened yet
Two reasons. First, the standards needed to make it real — W3C DIDs and Verifiable Credentials 2.0 — only reached Recommendation status in 2022. Until then, "open identity" meant "competing whitepapers." The bricks were not baked.
Second, the incentive layer was wrong. Self-sovereign identity projects in the 2018-2022 era (Sovrin, uPort, others) built credible architectures and ran out of runway. Identity protocols are platform-shaped: network effects take years; revenue is gated on adoption you cannot manufacture; smart teams quit before the curve bends.
What changes in 2026 is the convergence of three forces. eIDAS 2.0 mandates EU digital identity wallets for 450 million citizens by December. The agent transaction stack is consolidating around ERC-8004 (an agent identity registry) and x402 (credential-aware payment headers). The W3C standards floor is set.
These don't converge again. The window for the open layer to be built is now.
What Solidus is
We start with the wedge that pays: portable verification for humans, on a chain that anchors identity claims rather than financial transactions. We use the W3C standards as-shipped — did:solidus is a method submitted to the W3C DID Method Registry (under review, not merged, never "registered"). We use SD-JWT VC and BBS+ for selective disclosure, both production-ready, both shipped on our testnet.
The same primitive extends upward as the agent stack and the machine stack mature. We are not building a separate identity protocol for agents and another for machines. The same did:solidus method, the same VC format, the same selective disclosure path.
That is the convergence: one credential primitive, three entity types, every relying party.
What this is not
It is not a replacement for Auth0, Sumsub, Privy, or Skyfire — they each own a slice of the market, and our job is to be the layer underneath them, not the head-to-head competitor. We are betting the layer position is where the leverage is, the way TCP/IP was the leverage position underneath the SaaS stack of the 1990s.
It is not a token-launch event. There is no SLDS allocation before mainnet. The pitch is the protocol and the shipped components, not a speculation mechanism. If you are evaluating this for a token-rotation strategy, this is not that.
It is not a finished product. Eight of nineteen planned surfaces are live on testnet today. The other eleven wait on proof of the first eight, not on capital.
What you can check today
The SDK is on npm — @solidus-network/sdk covers DID method, VC format, BBS+ proofs, chain client, wallet interactions. The repo is open at github.com/solidusnetwork. The W3C DID Method Registry PR is #713. We are a DIF Associate Member, contributing to the Identifiers and BBS+ working groups.
If you build on identity, build on the standards underneath. If the standards underneath need a chain that anchors them honestly, that is the bet.
Top comments (0)