How Meta Engineered Ultra-Narrow Batteries for AI Glasses

· Source: Engineering at Meta · Field: Technology & Digital — Robotics & Autonomous Systems, Emerging Technologies & Innovation, Artificial Intelligence & Machine Learning · Depth: Intermediate, short

Summary

Meta engineered ultra-narrow steel-can batteries for its AI glasses, including Ray-Ban Meta and Oakley Meta Vanguards, to power features like cameras, speakers, and AI workloads within temple arms narrower than a pinky finger. Traditional pouch cells were inadequate due to their shape, volume inefficiency, and difficulty providing peak power. Meta's solution involved developing steel-can cells as narrow as 7mm, a width previously unachieved. This required redesigning internal components, such as replacing the "jelly roll" electrode architecture with die-cut stacked layers to achieve dramatically lower impedance. Tighter tolerances, around 100 microns, also contributed to increased energy density. Capacity evolved from 160 mAh in Gen 1 Ray-Ban Meta to 210 mAh in Gen 2, with system-level efficiencies doubling runtime. The Meta Ray-Ban Display glasses feature a 248 mAh cell for sustained power. This adaptable technology is now being scaled across Meta's hardware portfolio and multiple vendors.

Key takeaway

For wearables engineers designing compact AI devices, you must rethink battery form factors beyond traditional pouch cells. Consider ultra-narrow steel-can designs, like Meta's 7mm cells, which use die-cut stacked electrodes for lower impedance and leverage tight tolerances for energy density. If your design incorporates multiple batteries, plan for complex system-level challenges such as cross-charging risks and boot/shutdown sequencing. Prioritize system-wide power management and firmware control to maximize runtime, as chemistry alone may not suffice.

Key insights

Meta innovated ultra-narrow steel-can batteries with novel electrode architecture and tight tolerances for AI glasses.

Principles

Method

Replaced traditional "jelly roll" electrode with die-cut stacked layers for lower impedance and higher peak power delivery.

In practice

Topics

Best for: AI Hardware Engineer, Robotics Engineer, AI Engineer

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Engineering at Meta.