Ayar Labs’ UCIe-Compatible ‘TeraPHY’ Optical I/O Engine Extends AI Scale-Up Beyond the Rack, Strengthening Ecosystem Partnerships

August 9, 2026

TechArena Forum USA

Overview

Ayar Labs announced its “TeraPHY” optical I/O engine, built on silicon photonics and standard CMOS, connects via a UCIe-based electrical interface to enable AI scale-up beyond single-rack constraints. The company is forging partnerships with foundries, advanced packaging providers, OSATs, ASIC design services, and ODMs to develop comprehensive reference designs and manufacturing flows at the package, board, and rack levels. Ayar Labs is also actively recruiting talent to join its team transforming AI infrastructure with optical I/O technology, accelerating its growth strategy.

In Depth

Key Findings

Ayar Labs has developed its innovative “TeraPHY” optical I/O engine to address data movement bottlenecks in AI infrastructure. This engine, built upon silicon photonics and standard CMOS manufacturing processes, connects via a UCIe (Universal Chiplet Interconnect Express)-based electrical interface, enabling AI scale-up to extend beyond the physical constraints of a single rack. To facilitate broad deployment of this technology, the company is establishing strategic partnerships with foundries, advanced packaging providers, OSATs (Outsourced Semiconductor Assembly and Test), ASIC design services, and ODMs (Original Design Manufacturers), advancing the development of reference designs and manufacturing flows at the package, board, and rack levels.

Technical & Business Details

Innovation of TeraPHY Optical I/O Engine: TeraPHY achieves significantly higher bandwidth density and power efficiency compared to traditional electrical signal transmission by utilizing optical signals within or between adjacent chip packages. Compatibility with the UCIe standard ensures interoperability between chiplets from different vendors, greatly enhancing flexibility in AI system design. This dramatically boosts data transfer speeds between AI accelerators, maximizing the computational power required for training and inference of large-scale AI models.
Building an Extensive Ecosystem: Ayar Labs is not merely focused on developing individual technologies; it is dedicated to building an ecosystem that promotes industry-wide adoption. Collaboration with foundries optimizes TeraPHY’s manufacturing process to fit standard semiconductor production lines, ensuring mass manufacturability. Partnerships with packaging providers address chiplet integration and packaging challenges, while OSAT and ODM collaborations enable rapid integration into final products and faster time-to-market.
SuperNova Light Source: Ayar Labs is also developing the “SuperNova” light source, which works in conjunction with the TeraPHY chiplet. SuperNova provides highly efficient and stable optical output, serving as the foundation for transmitting data as optical signals through TeraPHY. This integrated approach offers an end-to-end optical interconnect solution, further optimizing AI workload performance.

Background & Context

AI workloads are pushing the bandwidth, power, and distance limits of traditional electrical connections for data movement between GPUs and AI accelerators. This “data movement bottleneck” is a primary factor hindering the performance scale-up of AI systems. Silicon photonics technology, which combines the advantages of optical communication (high speed, low loss, high bandwidth) with semiconductor manufacturing techniques, is considered the most promising approach to resolve this challenge. Companies like Ayar Labs are at the forefront of this field, aiming for a fundamental transformation of AI infrastructure.

Strategic Significance & Outlook

The widespread adoption of Ayar Labs’ TeraPHY optical I/O engine has the potential to revolutionize AI data center design and performance. The ability to scale AI beyond rack physical constraints will enable larger, more distributed AI systems, significantly reducing power consumption while maximizing computational power. The company’s ecosystem collaboration strategy will accelerate the standardization and proliferation of its technology, playing a crucial role in establishing the foundation for future AI computing. Ayar Labs remains a company to watch as a pioneer solving one of the most challenging computing problems in the AI era.

Source: https://techarena.ai/forum

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.