At Google I/O 2025, the company unveiled Google Beam, its next-generation video communication platform—and it’s a clear signal that Google’s AI-first strategy isn’t just about search, coding, or image generation. It’s about reinventing how we connect.

Beam is the commercial evolution of Project Starline, Google’s experimental telepresence system that aimed to make video calls feel like real conversations across a table. Beam takes that concept and pushes it into reality, powered by new breakthroughs in real-time AI modeling and a custom camera array.

Here’s what you need to know: Beam uses six camera angles to capture you from all sides. A new AI video model fuses these into a volumetric 3D projection—complete with near-perfect head tracking at 60 frames per second, in real time. The effect? You’re not just on a call. You’re present.

The End of Flat Calls?

If Beam delivers what it promises, it could mark the beginning of the end for traditional 2D video conferencing. Flat Zoom windows and laggy Google Meet grids aren’t built for natural interaction—they’re hacks, repurposing camera feeds into conversation simulacra. Beam wants to move past that.

By generating a live 3D model of the person on the other end, Beam restores physical presence—something even the best video calls can’t do. The subtle shifts of posture, the tilt of a head, the feeling of being in the same room—these are what Beam is aiming to capture. And early demos suggest it’s close.

Google says the first Beam devices will roll out to early adopters later this year, making it one of the first fully realized commercial applications of real-time volumetric video AI.

Why Now?

Beam’s arrival is perfectly timed. AI models have advanced to the point where real-time 3D reconstruction is not just possible but scalable. The compute power exists. The use case is obvious. And post-pandemic remote work culture is now permanent for millions of professionals.

Beam is Google’s attempt to claim the next frontier: presence. Not just communication, but spatial awareness and realism. It’s also a shot across the bow to Apple’s Vision Pro and Meta’s Horizon push—except Beam is less about avatars and more about real people, rendered through light and AI.

What’s Next?

Beam is still early. Questions remain about hardware costs, bandwidth requirements, and whether everyday users want—or need—this level of fidelity in their calls. But the ambition is clear. Google wants to make telepresence not just a feature, but a platform.

If it succeeds, Beam could be remembered as the moment we stopped talking on screens—and started talking through them.