Read the portfolio, not the patent. Looking at AR grants one at a time misses the pattern; aggregating them by CPC reveals it. The 2022 AR-optics grants pile into G02B 27/0172 — the head-mounted-display optics class — with Magic Leap holding a dense run including US11256093B2 (waveguide illuminator), US11320657B2 (waveguide display with cantilevered scanner), and US11422374B2 (focal planes for VR/AR).

The concentration in one subclass is the signal. When the meaningful patents in a category cluster this tightly in G02B 27/0172 rather than spreading across application-layer classes, it tells you the defensible work — and therefore the competitive fight — is in the optics, not the software. The light path is the moat.

“An optical system for an augmented reality head mounted display eyepiece that is configured to deliver images to the eye wherein the optical system includes optics. The optics are disposed so as to receive light output from the light source.”— U.S. Patent No. 11,256,093 source

Mapping the assignees in the cluster shows who owns the moat. Magic Leap and Meta (via its Reality Labs / Facebook Technologies entities) dominate, with optics licensors like Lumus and Dispelix and the occasional Apple, Snap, and Microsoft grant filling in. That is a concentrated field — a handful of well-funded players holding most of the issued optics IP.

Distinguish a filing surge from a grant surge. These are grants, reflecting filings from years earlier, so the cluster maps where the smart money was already investing well before 2022. A grant cluster is a lagging indicator of strategy — which makes it a reliable one for inferring sustained intent.

White space is the absence the map reveals. Where a category's grants concentrate this heavily in optics, the white space is rarely in the optics themselves — it is in the surrounding system: power, thermals, content, interaction. A new entrant is unlikely to out-file Magic Leap in waveguides, so the strategic opening is usually elsewhere in the stack.

For a strategist, the landscape lesson is concentration. The AR-optics IP is held by few hands and clustered in one subclass, which raises the cost of entry through the front door — and makes licensing, or competing on the parts of the stack the cluster does not cover, the realistic paths.