Teradyne has introduced Iris 100, a production-ready optical test platform for microLED devices, as manufacturers move the technology from laboratory development toward volume production. The platform is designed to test individual emitters across arrays containing millions of microLEDs, with applications spanning AR displays and optical interconnects being developed for AI data centers.
The race to build larger AI systems is creating demand for faster optical interconnects, but scaling those technologies requires more than new photonic components. Manufacturers also need production test systems capable of characterizing enormous numbers of individual emitters without turning optical testing into a manufacturing bottleneck.
That is the market Teradyne is targeting with Iris 100, a new optical test platform designed for high-volume microLED manufacturing.
The system combines a spectrometer and high-resolution camera to measure spectral response, per-pixel luminance and uniformity. Teradyne says parallel test and image-processing capabilities allow complete arrays containing millions of microLEDs to be characterized at production throughput.
The significance extends beyond display manufacturing. MicroLEDs are increasingly being explored for optical data interconnects, where large arrays of microscopic emitters can transmit data between computing and networking components. For AI infrastructure, that creates a potential path toward higher-density optical connectivity as electrical interconnects face increasing bandwidth, power and signal-integrity constraints.
Teradyne’s Iris 100 is designed to identify defects such as dead or stuck pixels and cluster defects earlier in the manufacturing process. Its NIST-traceable calibration workflow is intended to provide absolute measurements that can be correlated across lots, wafers and manufacturing sites, according to the company.
That manufacturing consistency becomes particularly important for optical AI infrastructure. A microLED array used in an optical interconnect is not simply a display component: variations across individual emitters can affect optical performance across an entire link. Testing therefore becomes part of the process of establishing reliable photonic devices at scale.
Iris 100 also integrates with Teradyne’s UltraFLEXplus automated test platform, allowing manufacturers to combine optical and electrical testing. The company says the system supports both wafer and final test, potentially allowing microLED characterization to be incorporated into existing semiconductor production workflows.
This is part of a broader expansion of Teradyne’s photonic test capabilities. Earlier in 2026, the company introduced Photon 100 for testing silicon photonics and co-packaged optics. Teradyne also acquired photonic test specialist Quantifi Photonics in 2025, expanding its portfolio across optical measurement and validation.
The strategy reflects a broader change in AI hardware manufacturing. Optical technologies are moving closer to processors, accelerators and switches, with silicon photonics, VCSELs, co-packaged optics (CPO), near-packaged optics (NPO) and optical chip-to-chip links all being developed to address the bandwidth requirements of AI systems.
Testing those components at production scale is becoming increasingly important because AI infrastructure is no longer being built primarily around isolated accelerator chips. Systems increasingly combine GPUs, CPUs, high-bandwidth memory, networking silicon and optical components into tightly integrated architectures.
The investment behind that transition is substantial. IDC estimates that global AI infrastructure spending reached $318 billion in 2025 and forecasts $487 billion for 2026, with spending expected to exceed $1 trillion by 2029.
Networking is already reflecting that expansion. IDC reported that the data-center Ethernet switching market grew 64.5% year over year to $12.3 billion in the second quarter of 2026, while 800GbE represented 41.2% of data-center Ethernet revenue.
As AI clusters push toward higher-speed networking, the components connecting compute resources must scale alongside them. That creates opportunities not only for optical engine and photonics suppliers such as Lumentum, Coherent and Broadcom, but also for companies providing the manufacturing and test infrastructure required to qualify those technologies.
Teradyne’s approach is consequently broader than a microLED display application. By bringing automated optical characterization into semiconductor production environments, Iris 100 addresses one of the less visible challenges in scaling photonic hardware: proving that millions of individual emitters perform consistently enough for commercial systems.
The platform does not by itself establish that microLED optical interconnects are ready to replace conventional electrical connectivity across AI data centers. Instead, it represents an infrastructure layer that could help manufacturers move promising optical technologies from prototypes toward repeatable, high-volume production.
For AI hardware developers, that distinction matters. As optical interconnects become more tightly integrated with compute, the ability to manufacture, calibrate and test photonic devices at semiconductor-like scale could become as important as the optical architecture itself.
Market Landscape
AI infrastructure growth is increasing pressure on the entire hardware supply chain, from accelerators and HBM to networking silicon, optical engines and semiconductor test equipment. IDC projects AI infrastructure spending to reach $487 billion in 2026, up approximately 53% year over year, and exceed $1 trillion by 2029.
The networking layer is scaling rapidly as well. Data-center Ethernet switching revenue grew 64.5% year over year in Q2 2026, reaching $12.3 billion, according to IDC.
This environment is encouraging investment in photonic technologies that can move data more efficiently, while creating a parallel requirement for automated optical testing. Teradyne’s Iris 100 sits at that manufacturing layer, alongside test technologies for silicon photonics, CPO and other high-speed optical components.
Top Insights
- Teradyne Iris 100 targets high-volume microLED testing for AR displays and emerging optical interconnects in AI data centers.
- The platform uses spectroscopy, imaging, parallel testing and software algorithms to characterize arrays containing millions of individual emitters.
- Integration with UltraFLEXplus brings optical and electrical testing into an existing semiconductor automated-test workflow.
- Teradyne’s photonics expansion includes Photon 100 and its 2025 acquisition of Quantifi Photonics.
- Production-grade optical testing could become increasingly important as photonic components move closer to AI compute and networking silicon.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
