Process compatibility and cumulative yield are overtaking peak brightness and pixel size as the key measures of commercialization

Research and Analysis by UBI Research

Reviewed by Dr. Choong Hoon YI, CEO and Chief Analyst

Competition in MicroLED microdisplays is shifting from peak brightness and pixel size to CMOS bonding yield and fab economics. As pixel pitch moves below 5 μm, LED chip performance alone is no longer sufficient. Commercial viability increasingly depends on integrating epitaxy, chip fabrication, transfer, bonding, color conversion, driving and inspection into one manufacturing system.

This shift was a central theme at the MicroLED Connect 2026 and AR VR Connect masterclasses held in Eindhoven, the Netherlands, on September 15 by the MicroLED Association and TechBlick. The official program covered Fraunhofer’s LLO and LIFT transfer processes with quantum-dot color conversion, Veeco’s high-volume MOCVD technology, ASML’s AR waveguide lithography and Ruisong’s discussion of yield and manufacturing for MicroLED microdisplays.

Ruisong argued that CMOS bonding at pitches below 5 μm cannot be treated as a simple back-end step added after LED fabrication. The LED structure, CMOS backplane, bonding method and inspection process must be considered together from the initial design stage.

Conventional 4- to 6-inch III-V fabs face limits in precision and process capability for ultrafine pixels. Moving to 200 mm or 300 mm silicon CMOS fabs may improve precision and production scale, but it can also create heavy capital costs and low-utilization risk in the early market. These statements reflect the program and a speaker’s technical position; they do not establish verified mass-production yields or customer approval for any supplier.

Early MicroLED competition focused on high brightness, long lifetime and smaller chips. As chips shrink, however, sidewall damage can reduce efficiency while transfer accuracy, bonding precision and defect inspection become more demanding. Small variations created upstream can be amplified during bonding and color conversion, making cumulative yield more important than the best result from any single process.

MicroLED suppliers therefore need to be evaluated not only by brightness or PPI but also by wafer size, bonding architecture, color-conversion yield, inspection throughput and foundry compatibility. Suppliers of low-temperature bonding, precision alignment and metrology compatible with existing CMOS lines could gain importance. The ability to connect process data from MOCVD through bonding, color conversion and inspection may also become a major differentiator.

UBI Research believes effective yield and total cost should be assessed before peak brightness when evaluating commercialization. Quoting chip yield, bonding yield and color-conversion yield separately can overstate the real finished-product yield. The cumulative yield from wafer input to optical-module shipment should be distinguished from yield after repair.

Use of 200 mm or 300 mm CMOS wafers does not automatically guarantee lower cost. If early AR demand is limited, depreciation and low fab utilization can dominate manufacturing cost. The usable portion of existing foundry processes, the extent of process modification, bonded wafer area and the reuse of customer designs will be key investment variables.

Future industry announcements are likely to be judged less by a sub-5 μm prototype itself and more by wafer-level bonding yield, color uniformity, inspection and repair time, and customer evaluation. Verified compatibility with existing CMOS foundries could rapidly define the equipment and materials supply chain. If dedicated fab investment comes first while AR demand develops slowly, high fixed costs could delay the transition to mass production.

Sources