Ayar Labs and YieldWerx Automate Silicon Photonics Testing for AI

Bringing Chip-Scale Optics to the AI Era

As artificial intelligence workloads continue to explode, the semiconductor industry is racing to overcome one of its biggest bottlenecks: moving massive amounts of data quickly and efficiently between chips. Ayar Labs, a pioneer in silicon photonics for chip-to-chip connectivity, and YieldWerx, a provider of advanced semiconductor test data analytics, have announced a strategic partnership aimed at automating and streamlining how next-generation optical I/O chips are tested and qualified.

The collaboration centers on Ayar Labs’ SuperNova optical source and TeraPHY optical I/O chiplets, and YieldWerx’s cloud-based YieldWerx platform for test data management and analytics. Together, they aim to create a more scalable, automated test infrastructure that can support the high-volume manufacturing needed for AI, high-performance computing (HPC), and data center applications.

Why Silicon Photonics Matters for AI and Data Centers

Traditional copper-based interconnects are increasingly struggling to keep up with the bandwidth and energy efficiency demands of modern AI training and inference systems. Large language models, recommendation engines, and other compute-intensive workloads require:

Terabits-per-second (Tbps) bandwidth between chips and accelerators
Lower latency to avoid bottlenecks in distributed training
Reduced power consumption to keep data center energy costs and heat under control

Silicon photonics – the integration of optical components directly onto silicon chips – offers a path to meet these demands. By using light instead of electrical signals for data transfer, optical I/O can deliver:

Higher bandwidth density
Longer reach without signal integrity loss
Better energy efficiency per bit

Ayar Labs is one of the leading companies pushing this technology from R&D into commercial deployment. However, the shift from conventional electrical interfaces to optical ones introduces new test, calibration, and reliability challenges at the chip, module, and system levels.

The Challenge: Testing Complex Optical I/O at Scale

Optical I/O devices such as Ayar Labs’ TeraPHY chiplets combine:

Advanced CMOS logic
Integrated lasers and modulators
Photodetectors and waveguides

Unlike traditional chips, these devices must be tested not only for standard electrical performance but also for:

Optical power levels and losses
Wavelength stability
Signal integrity across optical links

Manual or semi-automated test flows can slow down production and drive up costs. As AI workloads scale, customers expect optical I/O solutions that can be manufactured and qualified with the same rigor and throughput as established semiconductor products.

How Ayar Labs and YieldWerx Plan to Automate the Process

The partnership between Ayar Labs and YieldWerx is designed to bridge this gap by combining cutting-edge optical silicon with data-driven test automation.

Key elements of the collaboration include:

Centralized test data infrastructure
YieldWerx’s platform will collect, store, and manage test data from Ayar Labs’ optical I/O devices across the full manufacturing lifecycle – from wafer sort and final test to system-level validation.
Advanced analytics and yield optimization
By applying analytics to high-volume test data, the system can identify patterns, correlations, and failure modes that help improve yield, reliability, and process control.
Automation of complex optical test flows
Optical performance tests that previously required custom scripts or manual intervention can be standardized and automated, reducing engineering overhead and improving repeatability.
Scalability for high-volume AI deployments
As hyperscalers and system vendors ramp up AI infrastructure, Ayar Labs can leverage a more robust and scalable test ecosystem, aligned with the needs of high-volume manufacturing.

Strategic Importance for the AI and HPC Ecosystem

The move toward optical I/O is part of a broader industry shift to overcome the limitations of traditional interconnects. Market analysts and technology roadmaps from organizations such as the Open Compute Project and leading cloud providers consistently highlight:

The growing importance of disaggregated and composable architectures
The need for co-packaged optics and optical chiplets
Rising demand for energy-efficient data movement as AI clusters scale

By focusing on the test and validation layer, Ayar Labs and YieldWerx are addressing a critical – but often overlooked – part of the path to commercialization. Robust, automated test solutions are essential to:

Guarantee long-term device reliability
Maintain consistent performance across large deployments
Shorten time-to-market for new generations of optical I/O

What This Means for Chipmakers and System Builders

For semiconductor manufacturers, outsourced semiconductor assembly and test (OSAT) providers, and system OEMs, the partnership signals:

Reduced integration risk: A more mature test and analytics framework makes it easier to adopt Ayar Labs’ optical chiplets into existing supply chains.
Better visibility into device performance: Centralized analytics enable faster feedback loops between design, manufacturing, and field performance.
Improved cost structure over time: As test flows are automated and optimized, per-device test time and cost can decline, supporting broader adoption.

Ultimately, this collaboration helps pave the way for optical I/O to move from niche deployments to mainstream infrastructure for AI and HPC clusters.

Conclusion: Building the Test Backbone for Optical AI Hardware

As AI models and data sets grow exponentially, the need for faster, more efficient data movement is shaping the next decade of semiconductor innovation. Silicon photonics and optical I/O are emerging as key technologies to break through bandwidth and power barriers. Yet these advances can only be realized at scale if they are backed by robust, automated, and data-driven test infrastructures.

The partnership between Ayar Labs and YieldWerx is a step toward that future. By combining optical chiplet innovation with sophisticated test data analytics, the two companies are helping to build the manufacturing and quality backbone that AI hardware will rely on. As the industry moves deeper into the AI era, such collaborations will be crucial in turning promising lab technologies into reliable, deployable systems powering data centers around the world.