Digital CMOS LiDAR technology and investment research

Technical bottleneck

• Bottleneck type: Capacity + Qualification + Manufacturing concentration
• Technical constraint: Pixel-level ToF accuracy and thermal stability. Each CMOS pixel must measure photon round-trip time to sub-nanosecond precision (~0.15 ns = 5cm range error). Thermal drift, dark current, and cross-pixel capacitance create measurement noise. No competitor has yet achieved >50M pixels per chip at >60 fps with <5 cm accuracy at 100+ meter range.
• Economic constraint: Fab capacity allocation. Digital CMOS LiDAR on 28nm nodes targets $500–$1,500 at volume (millions/year). TSMC/Samsung prioritize GPUs and automotive SoCs; LiDAR allocation is secondary. Queue times for 28nm wafers are 6–12 months, creating a supply bottleneck independent of chip technology readiness.

Adoption

• Driver: Robotics deployment acceleration (IFR 2025: 850K+ robot installations, 15–20% YoY growth). Every robot needs 1–4 LiDAR sensors. Defense/C-UAS adoption for drone threat detection ($2B+ DoD allocation 2025–2027). Cost reduction trajectory (digital CMOS promises 3–5x cost reduction vs analog over 3–5 years).
• Blocker: Fab capacity concentration and long qualification cycles. Switching LiDAR suppliers in robotics/AV takes 12–24 months. Reliability unproven at scale — field deployments >10,000 units are rare. Alternative perception (stereo + radar in some AV programs) may contract LiDAR TAM if it proves sufficient.

Product categories

Rotating 3D LiDAR (analog, proven) · Solid-state LiDAR (mixed analog/digital, emerging) · Digital CMOS LiDAR (digital, early scale) · Stereo depth cameras (vision-based, complementary) · Hybrid stacks (LiDAR + stereo, emerging standard)

Public companies exposed

OUST (Ouster — primary digital CMOS supplier), LAZR (Luminar — analog/flash LiDAR), AMBA (Ambarella — edge vision SoCs), INNOVIZ (IVIZ — solid-state automotive LiDAR), CGNX (Cognex — vision systems, indirect exposure)

• AVAV (AVAV)
• NVDA (NVDA)
• DRS (DRS)
• BZAI (BZAI)
• AXTI (AXTI)

Validation signals

Ouster revenue FY2025 $169M (+52% YoY) with 49% GM validates digital CMOS cost curve. Q1 2026 $48.6M (+48.9% YoY) and Q2 guidance $49.5–$52.5M confirm sustained growth. Robotics/industrial now >50% of revenue (up from <30% two years ago). Ouster's acquisition of Stereolabs validates hybrid perception (LiDAR + stereo) as table stakes. arxiv 2025–2026 papers on LiDAR-inertial odometry, multi-sensor fusion, and calibration show active research in robotics integration.

Invalidation signals

Stereo/radar substitution in AV (Tesla, Waymo-lite models). If AV market widely adopts stereo-only, LiDAR TAM shrinks. Hesai dominance if U.S. export restrictions lifted (3–5x cost advantage would collapse competitor margins). Solid-state LiDAR failure if cost/reliability targets not met. Fab capacity never materialize if TSMC/Samsung deprioritize advanced node LiDAR allocation.

Related product map rows

• CGNX — Emerging: SENSORS/PERCEPTION: Machine vision = perception bottleneck for physical AI; CGNX dominant in industrial machine vision; robotics/warehouse automation ramp · confidence low
• AMBA — Early: Sensors/Machine Perception → Edge AI vision SoC bottleneck for autonomous robotics, ADAS, drone perception · confidence low
• OUST — Emerging: Edge AI → sensor proliferation → autonomous-system deployment (digital LiDAR) · confidence low
• CGNX — 2D/3D machine vision systems, vision software (VisionPro/In-Sight), barcode readers (DataMan): · confidence high
• OUST — OS (REV7) digital rotating LiDAR + ES2 solid-state LiDAR: 3D point cloud generation via CMOS digital pixel array; 360-degree volumetric sensing · confidence medium

Sources / open questions

• Research status: partial
• Notes: Updated 2026-06-14 during OUST (Ouster Inc.) deep-dive. Technology note expanded from stub to partial status. Key finding: digital CMOS LiDAR is moving from lab to production-scale deployment; the bottleneck is now fab capacity, not technology readiness. Ouster's 50%+ revenue growth and expanding gross margin (10% FY2023 → 49% FY2025) validate that the cost reduction curve is real. Next validation gates: (1) Ouster ES2 solid-state revenue contribution by Q3 2026, (2) fab capacity allocation signals from TSMC/Samsung 2026–2027, (3) stereo-only AV adoption trends.

Research Update — 2026-07-26

_Source: AEVA Q1 2026 EX-99.1, AEVA FY2025 10-K, AEVA June 2026 424B5_

Technical readiness: scaling in automotive and commercial-vehicle deployments; still qualification-bound rather than commodity-deployed.

Key papers / sources:

• SEC Aeva FY2025 10-K — 296 issued patents and 127 pending patent applications; Atlas range to 500m; explicit competitor set includes ToF LiDAR, FMCW LiDAR, and internal OEM programs.
• SEC Aeva Q1 2026 EX-99.1 — record quarterly revenue $6.3M; production-intent Atlas sensors delivered to Daimler Truck; first Atlas Ultra sensors delivered to a top European passenger OEM.
• SEC Aeva June 2026 424B5 — 4,494,382 shares at $22.25; ~$94.4M net proceeds before expenses; confirms capital dependence even after commercial progress.

Bottleneck update: strengthened; the constraint is OEM qualification + silicon-photonics integration + supply chain capital, not sensor physics alone.

Alternative/substitute risk: ToF LiDAR, in-house OEM LiDAR programs, and other FMCW vendors remain the main substitutes.

Timeline signal: 2027-2028 remains the key commercialization window; production-intent shipments are a necessary but not sufficient step.

Adoption rate data: Q1 2026 revenue $6.3M (+85.9% YoY) and 31.0% gross margin show early adoption, but operating loss $35.1M shows AEVA remains in ramp mode.

Thesis impact: AEVA validates that digital/4D LiDAR is moving into production-intent deployments, but economic capture is still early and capital markets remain part of the path.

Related research

• Sensing & Perception Master Deep Dive — consolidated machine-vision / LiDAR / sensing deep-dive (2026-08-14).