The technical case for laser-based AR display is not speculative. Higher efficiency, greater compactness, higher achievable brightness, wider color gamut, and for holographic architectures, fundamental capabilities like vergence-accommodation conflict resolution and prescription correction are real and measurable advantages over LED-based approaches. The challenges — waveguide uniformity, speckle in LCOS architectures, cost at current volumes, laser source availability optimized for AR — are all engineering problems with identified solution paths, not fundamental physical barriers.

Such were some of the issues discussed at a panel organized by the AR Alliance’s Laser Display for AR committee at SID Display Week 2026. It brought together practitioners from four companies working on distinctly different laser-based optical engine architectures:

• Moderator, Barry Silverstein, Director, Center for eXtended Reality (CXR), University of Rochester
• Theo Marescaux, Founder & Chief Product Officer, Swave (holographic SLM)
• Brian Chen, GM Innovation Center, Appotronics (LCOS and LBS)
• Joerg Reitterer, Co-Founder and CTO, TriLite (LBS)
• Kevin Curtis, SVP Advanced Optics Innovation, Magic Leap (waveguide technology)

The resulting conversation made clear that the AR Alliance’s laser working group, now comprising close to 30 companies, is building institutional infrastructure that could meaningfully accelerate the field — if it executes well.

Why Laser at All?

The panel opened with a deceptively simple question: given that LED-based approaches are already in product and microLED is well-funded, why does laser matter? The answers converged quickly on three physical advantages that are difficult to engineer around.

First, wall-plug efficiency at system level. Reitterer from TriLite explained that because lasers produce highly polarized, narrowband light with a very small étendue the fraction of generated photons that can be collected and directed into an optical system is substantially higher than for LED or microLED sources. For LCOS-based engines specifically, Chen from Appotronics noted that because both LCOS and laser light are polarized, you eliminate the polarization conversion loss that afflicts LED-illuminated LCOS systems — effectively doubling usable light from the source, which translates directly into either lower power for the same brightness, or higher peak brightness at the same power budget.

Second, compactness. The small étendue of a single-mode laser enables coupler size input on the order of one millimeter for LBS systems, compared to 3.5 to 5mm typically required for microLED-illuminated waveguides. Curtis from Magic Leap noted that the polarization advantage of laser alone buys approximately 35% improved waveguide efficiency relative to an unpolarized source — before accounting for the étendue difference.

Third, brightness ceiling. At 25-degree field of view, Chen cited measured peak brightness in the range of 5,000 nits from the laser-LCOS architecture. Silverstein observed that achieving comparable brightness over larger field-of-view angles is increasingly difficult for LED-based systems — a constraint that becomes decisive as the industry targets 40, 50, or 70-degree FOV for the next generation of glasses.

The framing Silverstein introduced — lumens per watt per cubic centimeter as the composite figure of merit for an AR light engine — proved useful for the rest of the discussion. It captures the three-way trade-off (efficiency, power, volume) that defines viability in a glasses form factor where weight, battery life, and heat are all hard constraints.

Architecture Landscape: Three Different Bets

The panel covered three distinct laser-based architectures, each with different trade-off profiles and target applications.

Laser-LCOS is the most mature of the three. Chen described current LCOS-based engines achieving approximately 0.2 cc for a 20-degree FOV system — compact, and manufacturable with established processes. The key constraint is pixel pitch: current production systems operate at around 3.8-micrometer pixel pitch, with the state of the art pushing toward 2.5 micrometers. Silverstein noted that research presented at the same conference included a one-micron meta-surface-based LCOS pixel, an important development. A particularly interesting architecture Chen raised is a binocular split-LCOS configuration — a single engine supporting both eyes — which he argued could substantially reduce bill-of-materials cost for the dual-display requirement of AR glasses.

Laser Beam Scanning (LBS) takes a fundamentally different approach: rather than projecting from a fixed pixel array, a MEMS mirror scans a focused laser spot across the field in a timing-controlled raster. Reitterer argued that this architecture’s scaling behavior is uniquely favorable: engine size does not grow with field of view or resolution, because the pixel grid is painted in time rather than instantiated in space. He put a sub-0.5 cc figure on a 50-to-70-degree FOV LBS system with high brightness output. The LBS approach also enables foveation natively — adaptive resolution, refresh rate, and brightness within a region of interest — which matters both for perceptual quality and for computational efficiency in high-FOV systems. For multi-beam LBS (which Silverstein noted Meta had explored as a laser array approach for each color channel), the path to higher resolution involves combining multi-emitter arrays with foveation schemes rather than simply scaling the MEMS mirror or scan speed. MEMS components themselves, Reitterer noted, are manufactured at scale for smartphone sensor applications and are unlikely to be a cost bottleneck.

Holographic display via SLM is the most architecturally distinct approach. Marescaux from Swave described a spatial light modulator based on phase-change materials with 282-nanometer pixels fabricated on 22-nanometer CMOS process nodes. The sub-wavelength pixel pitch enables extreme beam steering angles, which in turn allows the SLM to be integrated directly onto a legacy diffractive waveguide in-coupler, without redesign while easing integration with existing diffractive waveguide architectures and reducing the need for waveguide redesign. The deeper capability of holographic display is wavefront shaping. By computing and driving the interference pattern of the optical field at the SLM, the system can place displayed objects at the correct focal depth, eliminating the vergence-accommodation conflict that causes visual fatigue in conventional stereoscopic AR systems. The holographic approach also enables dynamic prescription correction — the user’s Rx is dialed in as a software parameter, removing one lens element and contributing to weight reduction. Marescaux positioned Swave’s technology as best suited for all-day-wear, always-on informational AR: notifications, navigation, AI agent interaction, and digital overlay — what Silverstein called the “”always-on contextual AI/AR display” use case — with all-day AI/AR information display as the initial beachhead, rather than immersive gaming as the first target use case.

Speckle: Manageable, Not Eliminated

Speckle — the coherent interference artifact that produces a grainy appearance in laser-illuminated images — received substantial attention, with a notably pragmatic conclusion. Chen drew on Appotronics’ cinema laser projection background to frame the target: current LCOS-based systems achieve approximately 15% speckle contrast, with a target below 5%. Silverstein noted that this range is consistent with what his team achieved in IMAX laser projection systems — and if that level is acceptable in a cinema environment, it is likely acceptable in AR.

Reitterer made the architecturally important point that speckle is simply absent in well-designed LBS systems, because there is no optically rough surface in the beam path between the laser source and the waveguide. Speckle requires coherent light impinging on a surface with roughness larger than the optical wavelength. The clean optical path of an LBS system (laser, lenses, prisms, mirrors, scanning mirror, waveguide gratings and TIR surfaces) contains no such surface.

For laser-LCOS, speckle mitigation techniques include high-frequency modulation of the laser drive, vibrating or phase-distorting diffuser elements, and waveguide optical design — a multi-technique approach that the panel regarded as engineering work, not a fundamental barrier. Marescaux similarly drew on Barco cinema laser projector experience to argue that the problem is solvable across laser display architectures.

The Waveguide Interface: Still the Hard Problem

If there is a single technical theme that ran through the entire panel, it is that waveguide compatibility and uniformity — not the light engine itself — is the binding constraint for laser-based AR display systems.

Curtis from Magic Leap offered a historical perspective: Magic Leap has spent many years researching laser beam scanning. The limiting factor was not speckle, not form factor, and not efficiency. It was color uniformity in the waveguide. Narrowband laser sources represent a new problem for diffractive waveguides designed around LED sources: because the illumination spectrum is extremely narrow, the diffractive grating behaviors are more sensitive to manufacturing variation, and pupil replication. The result is intensity non-uniformities that are hard to manage. The toolbox for addressing this, Curtis noted, has developed significantly over the past nine years, and Magic Leap’s current research direction is toward laser-compatible waveguides that it intends to supply to the broader AR ecosystem.

Chen’s perspective from the engine side was complementary. Appotronics deliberately co-designs its optical engine output characteristics with waveguide partners rather than treating the two as independent problems. The final image quality is governed by the system, not by either component in isolation. The goal is to engineer light distribution from the engine that works with the waveguide’s pupil replication behavior — not to demand 100% uniformity from either side independently.

For LBS specifically, the beam walk-off problem arises when the MEMS scanning mirror moves the beam across the input coupler, causing varying coupling efficiency at different field angles. Reitterer described the mitigation: relay optics between the MEMS mirror and the input coupler can hold the beam stationary on the input coupling grating, with only the angle varying — the necessary approach for high-FOV systems, at the cost of some added optical complexity. For low-FOV systems, the beam walk-off is inherently limited, so relay optics are not required.

Curtis also flagged the ICG size difference between laser and microLED sources as a waveguide design implication: a one-millimeter ICG for LBS versus 3.5-5.0mm for microLED means substantially different waveguide architectures, and a laser-optimized waveguide can achieve significantly higher coupling efficiency. At 70-degree FOV, a dual-sided waveguide imprint is likely necessary to keep the waveguide form factor acceptable.

Cost and Scalability: Volume Is the Answer

The panel’s most practically important section addressed cost — and the conclusion was that volume is the primary lever for almost every component in the bill of materials.

Curtis articulated the industry target bluntly: Meta had set a $100 system cost target for a single eye complete display and projection system. Magic Leap’s current roadmap is built around reaching that cost point through scaling and purpose-designed low-cost technology. Imprint-based waveguide manufacturing — Magic Leap’s current scalability strategy — was described as a key enabler.

On the laser source side, the historical analogy is instructive. Curtis noted that DVD-era laser diodes reached approximately 25 cents per unit at volume. Chen’s analysis broke down the cost structure of single-mode lasers: the incremental cost versus a standard LED is essentially the laser cavity, and at sufficient volume, cavity costs are not structurally different from LED manufacturing costs. The comparison that matters for the AR market is against microLED arrays — and there, Chen argued, lasers should be able to undercut microLED on cost because microLED requires many manufacturing steps and achieving high yield at small pixel pitch remains difficult.

Swave’s approach to scalability starts at the imager level: fabricating the SLM on 22-nanometer CMOS process nodes means leveraging existing semiconductor fab capacity without requiring purpose-built display fabs. Marescaux described Swave’s positioning as a chipset vendor — supplying the SLM chip, a compute companion chip, and a reference design for the holographic optical combiner — allowing customers to integrate without needing to develop the photonics from scratch.

For LCOS-based systems, Chen identified polarizing optical elements as a current cost issue, but not a structural barrier: liquid crystal optic approaches (explored at Meta) and metasurface-based optical elements are credible paths to bringing polarization optic costs down once volume justifies the manufacturing investment. The challenge is the transition period before volume exists.

Application Segmentation: Not One Technology Wins

Silverstein pushed the panel toward explicit application segmentation — a productive framing given that different architectures have materially different trade-off profiles. The emerging picture:

Holography (Swave SLM approach) is best suited for all-day-wear, lightweight, prescription-compatible informational AR — the always-on overlay device for AI agents, navigation, messaging, and digital world annotation. Not a monitor replacement. The vergence-accommodation conflict solution and prescription-in-software capability are differentiating features for long wear sessions.

Laser-LCOS covers a wider range, from the AI glasses snackable-display use case (compact, good battery life) to high-resolution specialty applications requiring 100% APL display — photography review, detailed visualization — where LCOS’s ability to drive every pixel at full brightness is an advantage. The binocular split-engine architecture potentially enables a cost-competitive path for the mainstream glasses market.

Laser Beam Scanning (LBS) becomes increasingly dominant as field of view increases. At 50-70 degrees and above, the lumens-per-watt-per-cc figure of merit is difficult for any LED-based source to approach, and the architecture’s FOV-independent engine volume is a structural advantage. Foveation is native to the scanning architecture, making it the natural candidate for high-resolution, wide-FOV systems — including eventually retinal projection geometries. Multi-emitter arrays combined with foveation are the path to resolution scaling beyond what a single-beam system can achieve.

Waveguide technology is not tied to any single light engine architecture. Curtis’s framing of Magic Leap as a waveguide platform company — intending to supply laser-optimized waveguides to any engine developer — suggests the waveguide layer may evolve toward commodity infrastructure, with differentiation shifting to the engine and compute layers.

Implications for AR Glasses Developers

For companies building AR glasses — whether consumer, enterprise, or industrial — the panel’s message is that laser-based optical engines are not a future curiosity. They are an engineering roadmap item with a credible near-term path for specific applications, and a likely necessary technology for wide-FOV systems that need to be worn all day.

The near-term practical implication is that glasses developers targeting mainstream form factors and extended wear sessions should be evaluating laser-LCOS and holographic SLM architectures seriously alongside microLED, not as a subsequent generation. The efficiency advantage is real and measurable, the speckle problem is manageable with known techniques, and the cost trajectory, while currently unfavorable, is volume-dependent rather than structurally unfavorable.

For wide-FOV glasses — anything above roughly 40 degrees that needs to be compact and battery-efficient — LBS deserves evaluation as the primary architecture rather than a niche option. The lumen-per-watt-per-cc advantage in this regime is not marginal. For lower-FOV systems, competing technologies can also achieve a small form factor — but the combination of high brightness and low power consumption still makes LBS an ideal technology.

The binding dependency for all laser architectures is the waveguide. Glasses developers should not treat their waveguide design as independent of their light engine choice. The ICG size, pupil replication design, grating structures, and uniformity requirements are all tightly coupled to whether the source is a laser, LED, or microLED. Co-development with a waveguide partner that is actively addressing laser compatibility — rather than adapting a microLED-optimized design — will determine whether the performance advantages of laser translate into product.

Implications for Component Suppliers

For laser manufacturers, the message is volume commitment. The cost gap between current laser sources and the $100 system target is not a physics problem — it is a volume and amortization problem. The AR Alliance’s laser working group, with its standardization efforts around per-architecture laser specifications, creates a mechanism for laser suppliers to develop against a defined target rather than guessing. Companies that engage with the AR Alliance process now and commit to laser designs optimized for AR engine architectures, are positioned to capture a large cost-sensitive market as it scales.

For waveguide manufacturers, laser compatibility is no longer optional for companies that want to serve the full addressable market. The uniformity and pupil replication challenges for narrowband sources are solvable — Curtis stated that directly, and the toolbox has expanded enormously in the past decade — but they require deliberate engineering investment. Companies that develop laser-optimized waveguide designs in parallel with their LED-compatible products will have a significant advantage as engine developers converge on laser sources for higher-performance applications.

For MEMS mirror suppliers, the LBS opportunity is real and the cost structure is favorable. MEMS technology is already manufactured at scale for consumer electronics sensor applications, and the additional requirements for AR display (scan angle, frequency, low deformation) are well within the capability envelope of existing MEMS fabs. The barrier is design optimization for AR-specific parameters, not manufacturing process development.

For polarizing optic suppliers, the trajectory toward liquid crystal optics and metasurface elements puts current polarizing optic approaches under medium-term cost pressure. Suppliers invested in conventional polarizing optics for AR should develop roadmaps toward these next-generation approaches, or risk being displaced when volumes justify the alternative manufacturing infrastructure.

What the AR Alliance Should Do Next

The AR Alliance’s Laser Display for AR working group is doing genuinely useful work — convening the supply chain, developing laser specifications per architecture, and creating a forum for the companies that need to co-develop solutions. But the panel surfaces several areas where more structured action would accelerate the field.

Publish architecture-specific laser specifications. The effort that Meta is leading within the working group — pulling together laser specifications for different engine architectures — should be formalized and published as a working group document. Laser suppliers cannot design to target without it, and AR glasses developers cannot make informed architecture choices without knowing what sources are available and at what cost trajectory.

Establish a waveguide-laser co-development track. The waveguide uniformity and pupil replication problem is the field’s most critical technical gap. The Alliance should formalize a co-development structure that pairs waveguide manufacturers with engine developers to work on laser-compatible waveguide designs — not as bilateral proprietary agreements, but as a working group initiative that produces shared learning and, where possible, shared test methodologies and metrics.

Define the system-level figure of merit. Silverstein’s lumens-per-watt-per-cc metric was used informally during the panel, but the field lacks an agreed framework for comparing architectures at system level across FOV, power, volume, and image quality. The AR Alliance is well-positioned to develop and publish a standardized evaluation methodology — analogous to what display standards bodies have done for emissive display technologies. This gives glasses developers a consistent basis for architecture comparison.

Address the cost roadmap explicitly. The $100 system cost target is in the room, but the AR Alliance has not published a component-level cost roadmap that maps current costs, volume thresholds, and technology transitions (e.g., liquid crystal optics replacing conventional polarizing optics, imprint waveguides reaching cost parity) to a plausible timeline. Doing so would help align investment decisions across the supply chain and provide a shared basis for OEM procurement planning.

Engage with prescription and medical standards. Holographic and advanced SLM-based systems that offer dynamic prescription correction represent a category that intersects with medical device regulation in most markets. The AR Alliance should proactively engage with standards bodies to clarify the regulatory pathway — both to protect companies investing in this capability and to prevent the regulatory environment from inadvertently blocking a clinically valuable feature.

Bottom Line

The field is not ready to displace microLED for the first wave of AR glasses. But it is ready to run in parallel, particularly for applications where form factor and all-day wearability matter most, and it is the necessary architecture for wide-FOV systems that need to be genuinely compact and power-efficient.

The AR Alliance’s working group structure is an appropriate vehicle for the supply chain coordination that the field requires. The question is whether it will operate with enough specificity and urgency — in laser specifications, waveguide co-development, system-level metrics, and cost road mapping — to compress the timeline to volume. The panel made clear that the technology is capable of getting there. The institutional question is whether the industry will organize around it effectively enough to make it happen.