Introduction

In September 2026 at MicroLED and ARVR Connect, I gave a “Master Class” on the conference’s first day. I tried to cram about 2 hours of content into a 1-hour presentation. I started with the slides and transcript from my presentation, then reordered them slightly and added more information and images. In cases where the slide pretty much says everything, I won’t add too much in the text.

In Part 1, I covered what I see as the major AR Challenges facing companies in 2026.

In Part 2, I will discuss the trends I noticed in 2026 for both Reflective and Diffractive waveguides. It seems that one year Meta demos a concept or releases a product, and the next year the market is flooded with other companies doing something similar. In 2024, Meta Introduced Orion with Silicon Carbide waveguides that use “WaveOptics-Like” (two gratings rather than three gratings) diffractive waveguides, and the next year there are multiple companies discussing Silicon Carbide waveguides. In 2025, Meta introduced the Meta Ray-Ban Display glasses with a Lumus 2-D pupil-expanding reflective (aka Geometric) waveguide, and in 2026, multiple companies showed 2-D pupil-expanding reflective waveguides.

Reflective (aka Geometric) Waveguides 2-D and 1-D Pupil Expanding

One of the first things that stood out to me in 2026 was seeing many more reflective (aka geometric) waveguides. Compared to the more common diffractive waveguide, reflective waveguides are more optically efficient (typically by a factor of 3 to 7), offer much better color uniformity, have less inherent forward light projection (aka eye glow), and can produce a wider FOV without requiring exotic glass or silicon carbide. These technical advantages are drawing competition to Lumus, which has dominated the category for over 20 years.

Many believe cost and yield have been the biggest factors holding reflective waveguides back. Lumus has worked with partners to reduce costs. In the last year or so, multiple companies are showing both 1-D and 2D pupil-expanding reflective waveguide prototypes.

Lumus started with 1-D waveguides in the early 2000s and in 2021 introduced its first 2-D pupil-expand design (see Exclusive: Lumus Maximus 2K x 2K Per Eye, >3000 Nits, 50° FOV with Through-the-Optics Pictures). Since then, Lumus has introduced both smaller and larger 2-D designs.

2-D Pupil Expanding Reflective Waveguides

Lumus

Lumus has been used in many industrial, military, and medical headsets for years and then most famously in the Meta Ray-Ban Displays in 2025. In 2026, Lumus also showed a 70° 2-D pupil-expanding waveguide prototype (below left). Lumus has also talked about making thinner, less expensive 2-D reflective waveguides in the 30° FOV range (below right). Lumus has worked with Schott and Quanta for many years to manufacture Lumus’s designs. In September 2026, Lumus announced a manufacturing agreement with Crystal-Optech to support higher volumes with a China-based manufacturer.

Optinvent has developed its unique plastic-molded reflective waveguide for over a decade, and at SPIE AR/VR/MR 2026 it showed a 2-D pupil-expanding prototype. Optix first showed a 2-D reflective waveguide in 2025, and at SPIE AR/VR/MR 2026 they showed an improved version. Envision Photonics showed both a 2-D reflective waveguide and a Hybrid combined Diffractive and Reflective waveguide at AR/VR/MR 2026.

Optinvent

Optinvent has developed its unique plastic-molded reflective waveguide for over a decade, and at SPIE AR/VR/MR 2026 it showed a 2-D pupil-expanding prototype.

Optinvent has shown 1-D pupil-expanding waveguides since at least 2013, starting with an Orca 1 prototype. At AWE 2018 they I got a picture of their prototype at the time.

The new 2-D expand waveguide helps shrink the projector engine that drives the waveguide, with a slimmer design and better integration (see below). I’m a bit concerned about the size/area of the Vertical expansion reflectors and how far over this forces the exit grating. It ends up thicker overall than glass reflective waveguides with similar FOVs. The big potential advantage is that it potentially has much lower manufacturing cost because it is an all-molded process rather than thin stacks of precision glass.

The view through the Optinvent shows considerable variation horizontally as the camera (or eye) moves slightly, revealing imperfections in pupil replication. The two pictures below were taken with slightly different camera locations. The variation is greater than what I’ve seen with glass-based (such as Lumus) reflective waveguides. I don’t want to be too critical here, as this seems to be an early prototype or proof of concept.

Optix

I first saw the Optix 2-D waveguide at AWE in June 2025 and then a newer design at SPIE AR/VR/MR in January 2026. They made significant progress that year in both waveguide size/area and image quality in a few months. Their overall waveguide design looks somewhat similar to Lumus’s 2-D waveguide, but it was still considerably larger in frontal area and less refined (still very much a proof of concept).

Their current prototype uses a fairly large DLP projection engine, but the waveguide could also work with LCOS. They are claiming a 50° FOV, 6,000 nits (to the eye with DLP), and less than 0.3% forward leakage (“eye glow). As Lumus has demonstrated for years, reflective waveguides have much lower forward projection than typical Diffractive waveguides. Below is a picture I took through the waveguide. I took this picture on the show floor, and the out-of-focus lower-right corner is a photo issue, not a waveguide issue. The color uniformity, while not perfect, is vastly better than any diffractive waveguide I have seen with a larger-than-30° FOV, but still not as good as the Lumus waveguides I have seen, at least not as of AWE 2026. You will also notice some geometric distortion, which is common with all wide FOV waveguides (reflective and diffractive).

Canon

Canon showed a 2-D expanding 30° FOV glass reflective waveguide at AWE 2026. They were showing the waveguide driven by both a MicroLED (about 340×240 resolution) and a Micro-OLED display. They also demoed the same 2-D waveguide with an OLED display, but it was very dim, as OLEDs are generally significantly less bright than most MicroLEDs. Canon’s demos were clearly preliminary prototype proofs of concept. Canon didn’t say whose displays they were using. The MicroLED display was low-resolution, about 320×240 pixels.

Early optical prototypes often lack the coatings that reduce reflections, which can cause the haze seen around the text in the picture I took through the optic below.

It’s hard to say where Canon is going with this proof of concept. Canon is a large corporation with the optical expertise to make it a factor in reflective waveguides, but it’s unclear whether Canon is really focused on the AR market.

Envision Photonics

Envision Photonics showed a 2-D reflective waveguide, a hybrid diffractive-and-reflective waveguide, and a reflective 1-D waveguide at AR/VR/MR 2026. Everything they showed was a prototype, not a product.

The 2-D reflective prototype (below left) had a very tall vertical expansion area, making the overall waveguide impractically tall. It’s more a demonstration of capability than a final design. They showed a >55° FOV with a 1920×1080 display and claimed it could reach >2000 nits at the eye. Envision Photonics showed a “hybrid waveguide” with diffractive entrance and vertical expansion gratings and reflective horizontal and exit prisms. The Hybrid waveguide supported a 30° FOV.

Envision showed only dark video and cartoon content, making it impossible to evaluate image quality seriously. Below are views through their 2-D 55° pupil-expanding reflective waveguide (left) and 30° hybrid waveguide (right). I suspect the image quality is not up to that of Lumus or Optix, but it is hard to judge with such content.

1-D Reflective Waveguides with OLED Microdisplays

We are also seeing several OLED designs with 1-D pupil-expanding reflective waveguides. Brighter OLEDs make these combinations work reasonably well, at least indoors, and a few products and prototypes are emerging. As stated above, Envision demonstrated a prototype using a SIDTEK MicroLED (right).

INMO has introduced several of these Micro-OLEDs with 1-D reflective waveguide glasses, the latest being their Air3 at CES 2026 (below left). The downside of the 1-D reflective approach is that it requires much larger optics before the waveguide, as seen in the Air3 below. Envision Photonics has also been showing 1-D reflective prototypes (below right) with OLED microdisplays.

Below is a diagram showing how the INMO 1-D reflective waveguide works (left) and the INMO Air3 Technical Specifications (right)

Below is an image I took through the INMO Air3 optics. You will notice uneven brightness across the image horizontally. The camera may exaggerate the issue, but it’s still visible to the eye.

Below is a close-up of the upper-right part of the image, with the camera in slightly different positions to mimic the eye moving relative to the “slats/prism.” This behavior is similar to that discussed earlier for Optinvent.

LetinAR and Tooz – Older Reflective “Non-Waveguide” Technologies

Among the longer-established approaches, LetinAR uses a reflective method with one-dimensional expansion, but is not thought of as “waveguides”. They are much thicker than waveguides and use only one (in the case of LetinAR) or a few (in the case of Tooz) total internal reflections (TIRs). What are typically considered waveguides are thinner and use many TIRs. LetinAR uses a single TIR to guide light at an angle to a curved collimating mirror that moves the display focus and reflects the image into the pupil expansion mirrors. As brighter OLED microdisplays become available, LetinAR has reduced the reflectivity of the pupil-replicating “slats” to reduce disturbance to the view of the real world. Like the Optinvent shown earlier, LetinAR uses a molded 2-piece plastic “sandwich” process: the various mirrors on one part receive their reflective coatings, then the two parts are glued together.

The older Tooz design (shown below right) used a Fresnel mirror with a similar 2-part molded construction. It was monocular, and the Fresnel mirror sat outside the main line of sight because looking through it would cause substantial distortion.

Diffractive Waveguides

Most AR Waveguides today use diffraction gratings to direct the light. Many companies make them, and large companies such as Meta design their own waveguides. AR glasses manufacturer Vuzix designs and manufactures diffractive waveguides both for their own glasses and for external sales. In 2025, Vuzix bought Google’s waveguide R&D facility (see: Google XR Glasses Using Google’s Raxium MicroLEDs While Waveguide Lab Sold to Vuzix).

Historically, most used a three-grating structure with entrance, expansion, and exit gratings. About 10 years ago, WaveOptics started demonstrating a 2-grating design.

The slide (right) lists many of the issues and trade-offs diffractive waveguides face. For example, there are typically trade-offs between improving efficiency and improving color uniformity. A wider FOV often requires more expensive high-index glass, or even silicon carbide, which has an even higher index. A big issue people talk about these days is the “rainbow” effect, where external light, such as overhead lights, is directed to the eye by one of the diffraction gratings, which also splits it into a rainbow of colors.

A common problem for diffractive waveguides is color uniformity, and it tends to worsen as the FOV increases. Most waveguide-based AR glasses today use green-only MicroLEDs with small FOVs, so color uniformity isn’t an issue for these glasses.

Compared to most companies that sell waveguides, Applied Materials is a very large company (~$30B in sales) best known for supplying equipment, services, and software for semiconductor manufacturing. They have made a major push into developing diffractive waveguides. The slide below shows four slides from Applied Materials’ AR/VR/MR 2025 presentation.

Two Grating Diffractive Waveguides

A trend I’ve noticed in the past few years concerns the waveguide grating layouts for diffractive waveguides. In prior years, most diffractive waveguides used three gratings, except for WaveOptics.

I sometimes call this the WaveOptics-like design because they have been making it for about 10 years. It has an input grating and a combined expansion-and-output grating. That makes the design more compact and removes a reflection people dislike: the so-called rainbow reflection often associated with the triangular third grating.

Snap acquired WaveOptics in 2021, and this type of waveguide appears in the Snap Specs, announced in 2026. A growing percentage of newer designs use two gratings. Most famously, in 2025, Meta’s Lab Prototype Orion used two double-sided grating waveguides (see my series of articles on Orion). Most famously, Even Realities uses a two-grating (glass) waveguide made by SEEV (below left). Nimbo X1 was recently announced using a two-grating silicon carbide waveguide also made by SEEV (below right).

Vuzix, which for many years made three-grating designs, showed different two-grating diffractive waveguides at CES 2026 (below).

Vuzix also showed very wide-FOV diffractive waveguides at CES 2026 that seem to have only two gratings, but with a much different exit/Expansion grating configuration. Most 2-grating waveguides center the entrance grating relative to the exit/expansion grating, but this design places the entrance grating in the corner, similar to most 3-grating designs. The waveguide was used in a military-type headset which was part of a Vuzix and Collins Aerospace joint development.

Meanwhile, many lightweight, green-only waveguides are appearing, including those from Even Realities G2, Vuzix Z100, RayNeo iO, and Rokid. These are among the simplest display glasses, just one step above audio-only products.

The content has to be simple. Green-only displays aren’t suitable for photos, videos, or web browsing. These products generally use microLEDs and try to limit heat by illuminating very few pixels. The average pixel level may be around 5 percent or less: typically, 95 percent of the pixels are off, while the illuminated pixels are driven fairly hard. The content is almost binary, consisting of simple text, arrows, and shapes.

Silicon Carbide Waveguides

There has been considerable discussion of silicon carbide (SiC) that started with Meta Orion. I recently wrote a somewhat skeptical article about it (see: Silicon Carbide Waveguides Pros and Cons); after all, I am known as an industry skeptic. Companies that have publicly shown SiC diffractive include SEEV (see above), Goertek (top left below), Moldnano (top right below), LLVision (lower left below), and Meta Orion (lower right below).

Goertek’s and Moldnano’s SiC waveguides use three gratings (entrance, expansion, and exit), while SEEV, LLVision, and Meta Orion use a two-grating method. Meta’s Orion has the gratings on both sides of the waveguide (for more on the Orion waveguide.

For more about Silicon Carbide and 2-grating versus 3-grating diffractive waveguides, see Silicon Carbide Waveguides Pros and Cons and Meta Orion AR Glasses (Pt. 1 Waveguides)

Next Time

In the next article in this series on my 2026 in-review based on my MicroLED and AR/VR Connect Masterclass, I intend to cover the new reflective meta-lens/meta-surface optics from Fourier Optics and TDK.