Bluesky’s active user base is shrinking as its focus expands beyond the app

Bluesky's mobile app shed more than half its monthly active users between its late-2024 high and June 2026. That is not a post-election hangover. That is a trend line.

Similarweb puts the Q4 2024 quarterly average at 22.1 million MAU. Q2 2026: 10.7 million. A 52% drop. The year-over-year figures — down 27% in June, daily actives down 26% in July — only confirm the trajectory. The app kept shrinking long after the Musk-driven surge should have settled into a new normal. It didn't. It kept bleeding.

The registered-account counter still ticks upward. That metric now measures sign-ups, not usage. The gap between accounts created and accounts opened is where relevance goes to die.

Stickiness offers the only comfort. The daily-to-monthly ratio holds at 29%, matching Threads. The people who remain show up. But a loyal rump does not a network make. Network effects demand growth, or at least stability. Bluesky has neither.

Toni Schneider knows this. The new CEO does not talk about winning the app war. She talks about AT Proto. The protocol is the product now. Bluesky the app is merely the reference implementation — a showcase, not the strategy.

This shift explains the apparent indifference to the app's decline. BlackSky and Eurosky are sprouting on the same substrate. Skylight, a video-first AT Proto client, has found early traction. These projects expand the protocol's surface area without adding a single user to Bluesky proper. That is the bet: a federation of niche communities outperforms a single generalist square.

The private-data push reinforces the pivot. Encrypted, permissioned spaces attract users who fled the public square entirely. They do not want a better Twitter. They want a group chat that scales. If AT Proto delivers that, the protocol wins even if the flagship app withers.

X's mobile numbers tell a parallel story. MAU down 3% year-over-year. DAU down 7% to 123.7 million. Web traffic up 5% to 4.7 billion visits. The legacy platform is hollowing out on mobile — the very surface where Bluesky competes — while migrating engagement to the desktop web. That is not a vote of confidence in Musk's direction. It is platform fatigue wearing a different uniform.

Users are not flowing back to X. They are evaporating.

The industry keeps framing this as a horse race. Bluesky versus Threads versus X. That frame misses the structural break. The generalist microblog is a mature category. Its growth ceiling is low. Its retention mechanics are broken by algorithmic feeds that optimize for rage-retention over community cohesion.

AT Proto's architecture — portable identity, composable moderation, federated discovery — attacks a different problem. It asks what social looks like when the graph belongs to the user, not the platform. That question matters more than Bluesky's MAU chart.

But the protocol bet carries its own danger. Ecosystems need anchor tenants. If the reference app becomes a ghost town, developers lose the proof point that the stack works at scale. Schneider must keep Bluesky viable enough to validate the protocol, while investing the real capital in the layer above. That tension will define the next two years.

Attie, the AI research tool, looks like a side bet. It diversifies revenue. It does not grow the graph.

The numbers are clear. The app is shrinking. The protocol is spreading. The CEO has chosen her horizon. Whether the market follows the protocol or demands a living app — that is the only question that counts.