Skip to main content

Bluesky's $100M Series B and the Quiet Bet on AT Protocol as Identity Infrastructure

· 11 min read
Dora Noda
Dora Noda
Software Engineer

A WordPress veteran is now running the social network the crypto industry didn't ask for. On March 19, 2026, Bluesky disclosed a $100 million Series B led by Bain Capital Crypto — a round that quietly closed in April 2025 and was never announced — alongside news that founder Jay Graber had stepped into a Chief Innovation Officer role and handed the CEO chair to Toni Schneider, the operator who scaled Automattic and helped turn WordPress into the open-source plumbing behind 40% of the web.

If you squint, this is the most consequential decentralized-identity bet of the cycle. And almost nobody in crypto is talking about it.

The round that hid in plain sight for eleven months​

Bain Capital Crypto led the round. Alumni Ventures, Anthos Capital, Bloomberg Beta, the Knight Foundation, and True Ventures came in alongside. Total venture capital raised across all rounds is now north of $120 million. The headline number is normal-sized for a 43-million-user consumer social product. The interesting number is the user growth: Bluesky entered its Series A round at roughly 13 million accounts and exits the disclosure of its Series B at over 43 million — more than tripling in twelve months — while crossing five million daily active users.

The hire is the louder signal. Toni Schneider spent fifteen years at Automattic, the company that built WordPress.com on top of the open-source WordPress engine and proved that you could run a profitable company while the protocol you depend on remains free, forkable, and federated. That is exactly the playbook Bluesky has stated publicly: "Just as WordPress created an open-source infrastructure that now runs more than 40 percent of the web, Bluesky plans to run the open social web on atproto."

Hiring the person who actually executed that playbook is not a coincidence. It is the company telling investors what business they think they are in.

What the AT Protocol actually is, in one diagram-shaped paragraph​

Strip Bluesky the app from atproto the protocol and you find four moving parts. Personal Data Servers (PDS) hold a user's signed repository — every post, like, follow, and profile field — as a content-addressed Merkle structure. Decentralized Identifiers (DIDs), most commonly the PLC method, anchor the user's cryptographic signing keys and rotation keys, and they resolve handles like alice.bsky.social to a permanent identity that can move between hosts. Relays subscribe to thousands of PDSes and merge their event streams into a single firehose. App Views — Bluesky's microblogging app is one of them — index that firehose to compute likes, follower counts, search, and feed algorithms.

Two properties fall out of this structure that are not true of any other social protocol at scale.

The first is what the AT Protocol team calls credible exit. A user owns the keypair behind their DID. If their PDS host disappears tomorrow, the rotation key still controls the identity, the repository can be re-signed and uploaded to a new PDS as a CAR file, and the network stitches everything back together — same handle, same followers, same posts. The Bluesky team rebuilt the inbound migration path in late 2025 specifically so that a user who left bsky.social for a self-hosted PDS could return without losing data. The protocol is now standardizing in the IETF, with the working group charter published in January 2026.

The second is algorithmic choice. App Views are interchangeable. A user can switch their default feed from Bluesky's chronological default to a third-party feed generator — there are now over a thousand of them in active weekly use — without leaving the network or losing follows. Custom moderation services work the same way. The "platform" is a frontend; the "protocol" is the data layer underneath.

The growth chart that breaks the decentralized-social comparison​

The standard hot take is that decentralized social is a forever-loser category. The numbers from the past eighteen months disagree, but only for one platform.

  • Bluesky: ~43M users, 5M+ DAU, growing 174% year-over-year through 2025
  • Nostr: ~16M total accounts, ~780K DAU, concentrated in Bitcoin culture
  • Lens Protocol: 1.5M historical accounts, ~20K DAU
  • Farcaster: smaller user base than Bluesky, deepest crypto-native developer mindshare, the strongest monetization story (Frames, paid casts) but a fraction of the consumer scale

Mastodon is the architectural ancestor everyone forgets to credit. ActivityPub federation predates atproto by years and powers a slow-growing, civic-feeling network of around two million monthly active users. The honest read is that Mastodon proved the federation thesis worked socially, and Bluesky proved it worked at consumer scale by inverting one assumption: in ActivityPub, your identity is tied to your home server's domain, so leaving means starting over. In atproto, your identity is tied to a key you own, so leaving costs nothing.

Farcaster is the more interesting comparison because both networks were funded by serious crypto capital and built by people who understood what the other was doing. Farcaster's architectural bet is on hubs (Merkle Manufactory-controlled aggregators) and a Frames-driven monetization layer that turns every cast into a potential transaction surface. It is the better bet if you believe social will be primarily transactional. Bluesky's bet is on identity portability as the protocol's load-bearing primitive and on the assumption that consumer scale comes first, monetization second. Both can be right. Only one of them currently has 43 million users.

The agent angle nobody is discussing yet​

Here is where the WordPress analogy starts to feel small.

Two months before Bluesky disclosed its round, the Ethereum Foundation and Virtuals Protocol shipped ERC-8183 — a standard that lets autonomous AI agents post jobs, escrow payments, and settle work on-chain without a human in the loop. ERC-8183 sits on top of ERC-8004, which launched on Ethereum mainnet in January 2026 and provides an identity and reputation layer for agents. Coinbase's x402 launched a parallel agent-payments rail. By Q1 2026, on-chain AI agents accounted for an estimated 19% of DeFi volume, and the count of daily active on-chain agents crossed 250,000 with year-over-year growth above 400%.

Every one of those agents needs three things: a verifiable identity, a portable record of what it has done, and a way to be discovered by other agents and by humans. ERC-8004 handles the first. The second and third are unsolved at the protocol level. They look an awful lot like what AT Protocol already provides for humans.

Imagine the social graph that emerges if an agent can hold a DID, publish a signed repository describing the jobs it has completed and the attestations it has received, and be followed, evaluated, and discovered through the same App View infrastructure that surfaces human accounts. The reputation primitive that ERC-8004 provides on Ethereum becomes legible to a network of forty-three million humans who already use the same identity rails. Agent commerce gets a discovery surface that is not Twitter, not LinkedIn, and not a fragmented set of crypto-native dashboards.

This is not Bluesky's stated roadmap. Schneider has been clear in interviews that the near-term focus is consumer growth, premium subscriptions, and a relay-and-moderation-as-a-service revenue line modeled on WordPress.com hosting. But the architecture does not need to change for the agent-era use case to materialize. It only needs the existing primitives — DIDs, signed repositories, App Views — to be pointed at non-human accounts. The Bain Capital Crypto thesis is almost certainly aware of this; Bain's portfolio is heavy on agent infrastructure, and a "social graph layer for the agent economy" maps cleanly to the protocol-distribution playbook the firm has been running across crypto.

The monetization question that decides the next two years​

A protocol-first company makes money by capturing a fraction of the value that flows through the protocol it stewards. Automattic does this through paid hosting, premium plugins, and managed services that benefit from running the canonical WordPress instance. Bluesky has telegraphed an analogous stack: premium subscriptions for power users, custom domain handle services, and managed relay, indexing, and moderation services that third-party PDS hosts can pay for instead of building themselves.

The catch is timing. Bluesky has been publicly opposed to surveillance advertising and to algorithmic preference for paying users — both of which are exactly the levers Twitter pulled to monetize. That principled stance is also why the platform attracts the users that crypto-native infrastructure investors care about: developers, journalists, researchers, and increasingly the policy and finance professionals who left X. But it means the runway has to outlast the time it takes to build a freemium business at consumer scale. With $120 million in cumulative funding and a five-million-DAU base, the math works if subscription conversion lands above 1–2%.

The harder question is whether atproto can support more than one credible app at consumer scale. Skylight (a TikTok-like video app), Flashes (an Instagram alternative), and integrations from Flipboard already exist. If a competitor app on atproto becomes the dominant short-form video surface and another becomes the dominant photo surface, Bluesky the company captures only the slice of value that flows through services it operates — relay, moderation, identity hosting. That is genuinely the WordPress outcome: an enormous protocol with a healthy commercial steward that no longer owns the consumer relationship outright. Investors who underwrote a 10x social-app outcome will not love that. Investors who underwrote a protocol-distribution outcome will.

What the disclosure actually signals​

Three things are now true that were not obvious before March 19.

First, Bluesky has eleven months of confidence behind a $100 million round. The fact that the company sat on the disclosure for that long suggests Bain Capital Crypto and the Bluesky board wanted the leadership transition, the federation milestones, and the IETF standardization process all to be in place before going public with the capital story. That is patient capital behaving patiently.

Second, the WordPress analogy is now operational rather than aspirational. Toni Schneider running the company with Jay Graber as Chief Innovation Officer is a clean separation between the protocol-stewardship role and the commercial-operator role. It is the structure Automattic settled into after Matt Mullenweg's product-led founding phase, and it is the structure that tends to outlast individual leaders.

Third, the standardization push at IETF moves AT Protocol out of "interesting Bluesky internals" and into "infrastructure other companies can build on without asking permission." The window between standardization and meaningful third-party adoption is usually two to four years. If that window opens before agent-era identity becomes a settled stack on Ethereum, Bluesky has a credible shot at being the social and identity surface for a generation of post-Twitter consumer products and a generation of agent-economy infrastructure that has not yet been built.

The decentralized social thesis was given a funeral several times between 2018 and 2024. The Series B that Bluesky disclosed yesterday is not a rebuttal so much as a quiet announcement that the people writing checks against this thesis have already moved on to the next question — not whether open social protocols can reach escape velocity, but which application categories they reach first.

BlockEden.xyz operates production infrastructure across more than a dozen blockchain networks, including the identity and agent-payment rails that AT Protocol-adjacent applications increasingly depend on. Explore our API marketplace to build on protocols designed to outlast their first commercial steward.

Sources​

Share on Twitter