Meta Connect produced three stories this week and the flashiest one is the furthest away. Meta VR Glasses — 100 grams, $1,299.99 — ships in spring 2027. Hologram avatars on that headset ship then too.
The device that actually matters in the near term got the quietest announcement: Meta Ray-Ban Display began rolling out internationally on 23 September 2026, having been US-only since launch.
Why availability is the story
Because Ray-Ban Display is where Hologram lands first.
From the hands-on reporting this week: the Ray-Ban Display version of Hologram arrives in Early Access this autumn, while the VR Glasses version waits for spring 2027. And the Ray-Ban version is the one that impressed — a tester initially mistook a shoulders-up Hologram for a real selfie camera feed, whereas the VR version looked blocky, undersized, and broke when you walked toward it.
That ordering is the opposite of how these announcements get covered. The premium headset gets the headlines; the glasses get the working feature, eighteen months earlier, on hardware people can actually buy.
What the device is
Ray-Ban Display pairs a monocular display in the glasses with the Meta Neural Band — the sEMG wristband that reads muscle activity at the wrist and turns subtle finger movements into input, with haptic confirmation. We’ve covered the Neural Band’s research lineage before; it is the most interesting mainstream neural-interface product shipping.
The combination is deliberately modest compared to a headset: a small display for glanceable information, plus an input method that doesn’t require speaking aloud or waving your hands in public. That modesty is why it works.
What creators should take from it
Hologram is a 2D video problem on this device, and that’s why it succeeds. As we wrote earlier this week, Hologram isn’t volumetric — it’s a server-side generative diffusion model driven by your microphone audio, producing a 2D video stream for Ray-Ban Display or stereoscopic video for VR Glasses. A small, flat, head-and-shoulders video on a phone call is precisely the regime where generative video is convincing. The same technique fails in a room-scale headset because it’s a billboard.
So if you’re thinking about presence, telepresence or avatar work: the version that works is the constrained one. Not a body in your space — a face in a call.
International availability changes the audience maths. A US-only device is not a platform. Anything built for Ray-Ban Display — glanceable interfaces, audio-first experiences, Neural Band interaction — now has a plausible non-US user base, which is the difference between a demo and a product.
The interaction model is the durable part. Whatever happens to any specific Meta device, glanceable display plus sEMG wrist input plus voice is a distinct interaction grammar. It rules out precise pointing and long text entry and rewards short, confirmable, one-handed actions. That’s worth designing against now, because the VR Glasses use a related model — gaze and gesture, no controllers — and the $1M developer competition Meta opened this week explicitly requires hands-only interaction.