← Back to all notes
R

Roel Huis in ‘t Veld

Director, Kriya Materials · Week of November 16, 2026

Reflective waveguides are winning on physics; plastic is where the cost fight will be

Karl Guttag published the second part of his 2026 AR year-in-review on 5 November, and the headline for me is not the field-of-view numbers but the material shift underneath them. Reflective, or geometric, waveguides, the Lumus family, are now described as typically three to seven times more efficient than diffractive ones, with better colour uniformity and less forward light leakage, and the competitors arriving to chase Lumus are increasingly building them from moulded plastic rather than glass stacks. Optinvent and LetinAR mould; Lumus laminates glass with Schott and Quanta. Guttag's own words: the moulded route is "potentially" cheaper "because it is an all-molded process", and in the comments he calls LetinAR-style designs "likely significantly less expensive than glass waveguides". My view is that this is the moment a coatings supplier should pay attention, because a moulded plastic waveguide is only as good as its surfaces. A reflective waveguide is a stack of partially reflecting mirrors at a precise angle. Every one of them is a thin-film coating, and in a polymer you deposit those films on a substrate that outgasses, expands far more than glass with temperature, and scratches if you look at it. The forward leakage Optix claims to have pushed below 0.3 percent is a coating number, not a geometry number. Eye glow, which Guttag again flags for diffractive designs, is in the end also about how much light the outer surfaces send the wrong way. So the question customers will ask next year is not glass or plastic, but who can put an anti-reflective, hard, index-matched stack on a curved polymer part for a few euros, with yield. That is a different supplier base than the one that grinds glass. Guttag also reports what he saw, and it is early days. The Optinvent prototype is thicker than glass designs of similar field of view and shows horizontal brightness variation. Canon's 30-degree glass design is a proof of concept. Envision's 2-D design is impractically tall. The diffractive camp is not standing still either: two-grating designs, in Snap Specs, Even Realities, Vuzix and Meta's Orion, are the trend, while wider fields of view still drag in expensive high-index glass or silicon carbide. Meanwhile the smart glasses that will actually ship this quarter mostly have no waveguide at all. Warby Parker said in its Q3 results on 5 November that pre-orders for its first Intelligent Eyewear collection, built with Samsung and Google, open on 12 November, with deliveries from mid-December; Android Police reports these are camera-and-audio glasses without a display. The display tier is green-only microLED, where Guttag puts average pixel levels at 5 percent or less, which is why those glasses can be light. I have been in optics long enough to remember when plastic was going to replace glass in camera lenses. It did, but only once the coating problem was solved. Waveguides will go the same way, and the margin will sit in the nanometres on the surface, not in the mould. Sources KGOnTech (Karl Guttag), “AR in 2026 Year In Review Presentation (Part 2: Waveguide Trends)”, 5 November 2026. https://kguttag.com/2026/11/05/ar-in-2026-year-in-review-presentation-part-2-waveguide-trends/ Warby Parker (SEC 8-K, Q3 2026 results), “Warby Parker Inc. Q3 2026 earnings release”, 5 November 2026. https://www.sec.gov/Archives/edgar/data/0001504776/000150477626000022/warbyparkerincearningsrele.htm Android Police, “Android XR glasses are finally getting a pre-order date”, 6 November 2026. https://www.androidpolice.com/samsung-android-xr-glasses-pre-order-nov-12/ Vision Monday, “Warby Parker Reports Q3 Revenue Growth, Preps for Intelligent Eyewear Launch Next Week”, 6 November 2026. https://www.visionmonday.com/business/article/warby-parker-reports-q3-revenue-growth-preps-for-intelligent-eyewear-launch-next-week/