The distinction that decides whether a head-worn display can run always-on, and the one most AR display comparisons skip. It is a power-architecture question, not an image- quality one.
The mechanism
- Modulated (LCoS, DLP): a separate illumination source runs, and a spatial modulator subtracts light to form the image. A black pixel is light that was generated and then thrown away. Power is roughly independent of content.
- Emissive (MicroLED microdisplays, microOLED) and additive (laser beam scanning — the lasers simply are not switched on for a dark pixel): only lit pixels consume power. Power scales with content.
AR content is overwhelmingly sparse — notifications, a few lines of text, a glyph on an otherwise black field. So for an always-on device the gap between the two is not marginal; it is the difference between a device you wear all day and one you charge at lunch.
Why the current volume winner is not the endgame
LCoS leads today because it is mature, not because it is efficient: monolithic RGB microLED is blocked on red EQE at small pitch (see MicroLED microdisplays). That makes laser-LCoS a strong candidate to win 2030 volume and to be replaced afterwards.
The larger lever: deleting the combiner
Caveat that keeps this honest: microOLED is emissive but architecturally boxed into birdbath/prism optics — it cannot reach the ~1-10M nits a waveguide needs, and its Lambertian emission is étendue-inefficient for waveguide injection. Emissive is necessary, not sufficient.