OpenLight pushes photonic integration as AI networks drive optical scaling
Key Highlights
- Integrating lasers directly onto photonic chips enhances efficiency, power savings, and thermal management in high-speed optical systems.
- The transition from pluggable optics to co-packaged optics addresses bandwidth density and electrical interconnect challenges in large AI clusters.
- OpenLight offers a photonic design platform with a library of components, enabling custom PIC development using standard semiconductor design tools.
- The platform's approach mirrors fabless semiconductor models, facilitating scalable, cost-effective manufacturing of optical components.
- Beyond communications, photonic integration is expanding into automotive sensing, industrial, defense, medical, and quantum computing markets.
At OFC, OpenLight focused on how photonic integration, particularly integrating lasers directly onto photonic application-specific integrated circuits (PASICs), is becoming increasingly important as the optical industry scales to support AI infrastructure, higher bandwidth optics, and co-packaged optics (CPOs).
Integrating lasers directly onto the photonics chip reduces the number of discrete components, eliminates complex optical alignment processes, and enables more scalable manufacturing.
Rather than selling fixed optical devices, OpenLight provides customers with a photonic design platform and component library, enabling companies to design custom PICs using standard semiconductor design tools and foundry manufacturing.
“We give customers a library of components to design chips to their own specifications,” said OpenLight CEO Dr. Adam Carter. “They can simulate circuits using standard semiconductor design software, send the design to the foundry, and then start manufacturing wafers. What we’re really trying to do is bring silicon economies of scale and cost structures to the photonics world. We’re trying to siliconize optics.”
OpenLight’s Photonic Design Platform
Unlike traditional optical component vendors that sell finished devices, OpenLight operates more like a semiconductor platform company. The company provides a photonic process design kit (PDK) that includes a library of photonic components—lasers, modulators, waveguides, and other building blocks—that customers can use to design their own photonic application-specific integrated circuits.
Engineers design their photonic circuits using standard electronic design automation (EDA) tools from semiconductor software companies, simulate the optical and electrical performance, and then submit the design to a semiconductor foundry for wafer fabrication. This model mirrors the fabless semiconductor industry, allowing companies to design custom photonic chips without building their own manufacturing facilities.
The approach is intended to bring semiconductor-style scaling, cost structures, and manufacturing processes to photonics, enabling higher levels of integration and potentially lowering the cost of optical components as volumes increase. The platform also enables the direct integration of active components, such as lasers, onto the photonic chip, which improves efficiency, reduces packaging complexity, and supports higher-density optical interconnects needed for AI, high-performance computing, and next-generation networking.
Integrated lasers improve efficiency and power consumption
By integrating the laser directly on the chip, coupling efficiency can approach roughly 90%, reducing the need for higher drive voltages and lowering overall power consumption. This level of integration becomes increasingly important as optical interconnect speeds move toward 1.6T and 3.2T, and as power efficiency and thermal constraints become critical in AI data centers and high-performance computing systems.
While DSPs still consume the bulk of power in optical modules, reducing optical power consumption contributes to overall system efficiency and thermal management in high-density environments. Even incremental improvements at the optical layer can have a meaningful impact at the system level in large-scale AI deployments.
Pluggables vs. CPOs
Carter also addressed one of the most discussed topics at OFC 2026: the transition from pluggable optics to co-packaged optics (CPOs). “Pluggables are not going away,” he said. “But their role will evolve as bandwidth density requirements continue to increase.”
The primary limitations of pluggable optics are electrical signal integrity and connector density. As bandwidth per port increases and the number of electrical lanes grows, electrical interconnects become increasingly difficult to scale. CPOs are emerging as a solution, particularly for scale-up architectures inside large AI compute clusters where optical interconnects can reduce reliance on short-reach electrical links between switches and adjacent components.
While pluggable optics will likely remain dominant in scale-out network architectures, CPOs will find their way inside large compute clusters where density and power constraints are more severe. Platforms that integrate active components such as lasers directly onto the PIC are expected to play an important role in enabling these higher-density architectures.
The rapid growth of AI clusters is understandably accelerating the transition. As GPU clusters scale toward hundreds of thousands or even millions of GPUs, the number of electrical interconnects becomes impractical, pushing the industry toward optical interconnects inside systems rather than only between systems.
Markets beyond communications
Communications is currently the largest and fastest-growing market for photonic integration. Still, OpenLight is positioning its platform across multiple industries, including automotive sensing, industrial sensing, defense, medical devices, and emerging technologies such as quantum computing and AR/VR. “Fortunately, each of these markets is at a different stage of maturity, providing multiple growth opportunities over time,” said Carter.
OpenLight recently announced its first customer production orders through its foundry manufacturing process, a milestone that demonstrates the platform meets reliability and performance requirements for production deployment. This progression from design enablement to production orders reflects increasing adoption of the platform model.
As AI infrastructure continues to push optical interconnect density, power efficiency, and integration levels, photonic integration platforms may play a growing role in the design and manufacture of optical components.
For related articles, visit the Optical-Tech Topic Center.
For more information on high-speed transmission systems and suppliers, visit the Lightwave Buyer’s Guide.
To stay abreast of fiber network deployments, subscribe to Lightwave’s Service Providers and Datacom/Data Center newsletters.
About the Author
