MicroLED microdisplays are arrays of inorganic GaN light emitters at micron-scale pitch bonded to silicon drive circuits, and they are the leading candidate light engine for waveguide-based AR glasses, with full-colour versions forecast to reach mature commercialisation in 2027-2028 ref.
Summary
A microLED microdisplay (often called LEDoS, LED-on-silicon) is a dense two-dimensional array of inorganic light-emitting diodes, typically grown in gallium nitride, driven by a silicon CMOS backplane. Unlike a liquid-crystal-on-silicon (LCoS) panel, which modulates light from a separate illuminator, each pixel emits its own light. That matters for augmented reality: in a glasses form factor the light engine has to push an image through a diffractive or holographic waveguide that throws away most of the photons, so the panel needs extreme brightness from a package measured in cubic millimetres. The competing engine families in this niche are LCoS and laser scanning, all three of which sat inside the same remit at Meta Reality Labs, described as “the world’s smallest and most efficient visual imaging systems” ref.
The hard parts are pixel pitch, full colour and yield. JBD’s Roadrunner II, introduced in June 2026, is a full-colour microLED projector at 2.5 micron pixel pitch shipped with a monocular AR glasses development kit ref. Full colour is the specific bottleneck the industry has been working through: TrendForce points to two enabling breakthroughs and forecasts full-colour LEDoS entering mature commercialisation between 2027 and 2028 ref. On the manufacturing side, JBD completed an upgrade to a 12-inch reconstructed wafer platform in May 2026, framed as a step change in manufacturing efficiency and cost competitiveness for mass-production readiness ref.
The deciding parameters are therefore lumens per watt at the eye after waveguide losses, panel yield at micron pitch across a full wafer, and the cost of full-colour integration. A second, adjacent research thread treats micro-LEDs as communication emitters rather than display pixels: cubic GaN lacks the spontaneous internal polarisation of the conventional hexagonal phase and so promises higher modulation speed for short-range optical links, though c-GaN is metastable and its selective-area growth in silicon inverted pyramids is still being characterised at the TEM level.
For a reader meeting this for the first time: the physics is not in doubt, the devices exist and are being sampled, and the open question is entirely industrial economics plus whether laser-based engines leapfrog the category before it reaches volume.
Evidence base
- 2026-04-07: TrendForce forecasts full-colour LEDoS devices entering mature commercialisation between 2027 and 2028, tied to strong growth in AR smart glasses demand ref.
- 2026-05-15: JBD completed an upgrade to a 12-inch reconstructed wafer platform, described as a step change in manufacturing efficiency and cost competitiveness for mass-production readiness ref.
- 2026-06-01: JBD introduced Roadrunner II, a 2.5 micron full-colour microLED projector, with a monocular AR glasses development kit ref.
- 2026-06-15: Barry Silverstein, previously Senior Director and CTO of Optics and Display at Meta Reality Labs with a remit covering µLED, LCoS and laser scanning engines plus waveguide technologies, joined the University of Rochester in January 2026 to direct a new Center for Extended Reality; his public position is that next-generation displays move to laser-based engines ref.
- 2026-07-01: Selective-area OMVPE growth study finds polarity is critical to cubic GaN growth in silicon inverted pyramids; c-GaN is metastable and prior reports contradictory, but its lack of spontaneous internal polarisation promises higher speed than hexagonal GaN for micro-LED short-range communication.
Open questions
- What is the actual full-colour light-engine efficiency in lumens per watt at the eye after waveguide losses, and what is panel yield at 2.5 micron pitch on the 12-inch reconstructed wafer platform?
- Does the laser-engine argument advanced by the former Meta optics and display CTO win any major OEM design, and on what performance grounds?
- Beyond the single supplier visible in these sources, is there a credible second source for full-colour microLED microdisplays that a consumer OEM would accept?
- Can cubic GaN growth be made reproducible in inverted-pyramid templates at device scale, and does the predicted speed advantage over hexagonal GaN materialise in measured modulation bandwidth?
Assessment drafted 2026-08-31 from up to 10 KB sources using the technology-scorecard framework; scores are a draft read pending review.