Lightmatter and Cadence Partner to Accelerate Co-Packaged Optics for the Future of AI Infrastructure
Lightmatter, a pioneer in photonic computing and optical interconnect technology, has announced a strategic technical collaboration with Cadence, a global leader in electronic design automation (EDA) and silicon IP. The partnership aims to accelerate the development of co-packaged optics (CPO) solutions by integrating Cadence’s high-speed, silicon-proven SerDes intellectual property with Lightmatter’s Passage™ optical engine.
By aligning advanced-node CMOS technology with industry-standard packaging workflows, the two companies seek to make CPO solutions not just technologically viable but also scalable and manufacturing-ready. The collaboration represents a critical step toward building next-generation AI and high-performance computing (HPC) systems capable of supporting the massive data movement demands of modern AI workloads.
As AI models grow exponentially in size and complexity, the limitations of traditional electrical interconnects are becoming increasingly apparent. Lightmatter and Cadence’s partnership addresses this challenge head-on, offering a roadmap for integrating optical and electronic technologies at unprecedented levels of performance, efficiency, and scale.
The Data Bottleneck in AI: Why Optical Interconnects Matter
Artificial intelligence is fundamentally a data movement problem. Training and deploying large-scale AI models require massive amounts of data to flow between processors, accelerators, memory, and storage systems at extremely high speeds and low latency.
Historically, AI infrastructure has relied on copper-based electrical interconnects, pluggable optical modules, and near-packaged optics to move data across systems. While these approaches have served the industry well, they are increasingly constrained by physical and economic limits.
Electrical interconnects suffer from signal degradation, power inefficiency, and limited bandwidth as distances increase. Pluggable optics improve bandwidth but introduce latency, power overhead, and packaging complexity. Near-packaged optics reduce some of these limitations but still fall short of the scalability required for future AI clusters.
Co-packaged optics represents the next major architectural shift. By integrating optical engines directly with compute silicon within the same package, CPO dramatically reduces the distance data must travel electrically before being converted into optical signals. This approach delivers higher bandwidth density, lower power consumption, and improved system-level performance—key requirements for the next generation of AI and HPC systems.
Lightmatter and Cadence’s collaboration is positioned at the center of this transition.
Combining Strengths: Photonics Meets Silicon IP
At the core of the partnership is the integration of Cadence’s optics-optimized, high-speed SerDes IP with Lightmatter’s Passage™ optical engine. SerDes (serializer/deserializer) technology is essential for high-speed data transmission, converting parallel data into serial streams and vice versa. In modern AI systems, SerDes performance directly impacts bandwidth, latency, and energy efficiency.
Cadence brings decades of expertise in silicon IP, advanced-node CMOS design, and EDA tools to the collaboration. Its Universal Chiplet Interconnect Express™ (UCIe™) IP further enables disaggregated chiplet-based architectures, a key trend in modern semiconductor design.
Lightmatter, meanwhile, contributes deep expertise in silicon photonics, laser technology, and optical interconnects. Its Passage platform is designed to deliver scalable, high-performance optical connectivity tailored for AI and HPC workloads.
By combining these capabilities, the partnership aims to create a tightly integrated optical-electrical stack that can be adopted by hyperscalers and semiconductor companies building custom AI chips.
A Turning Point for AI Infrastructure Architecture
The shift toward co-packaged optics marks a fundamental change in how AI infrastructure is designed. Instead of treating optical connectivity as an external component, CPO integrates photonics directly into the compute package, enabling new levels of performance and efficiency.
This transition is often described as an inflection point for the industry. It moves AI systems beyond traditional pluggable optics and near-packaged designs toward fully integrated, 2D and 3D-stacked photonic architectures.
In such architectures, optical engines are co-located with processors and memory, reducing latency and power consumption while dramatically increasing bandwidth density. This approach is particularly important for hyperscale data centers, where energy efficiency and scalability are critical economic drivers.
Lightmatter and Cadence’s collaboration addresses not only the technological challenges of CPO but also the practical realities of manufacturing and deployment. By focusing on compatibility with advanced-node CMOS and industry-standard packaging workflows, the partnership aims to accelerate real-world adoption of CPO technologies.
Industry Voices on the Collaboration
Ritesh Jain, Senior Vice President of Engineering and Operations at Lightmatter, emphasized the importance of rethinking data movement in AI systems.
“The next big leap in AI performance requires a fundamental change in how we move data,” Jain said. “Cadence’s connectivity IP is an ideal complement to our Passage platform. Together, we are paving the way for CPO deployment by solving the most complex optics-electronics integration challenges, ensuring that the next generation of AI clusters can achieve the energy efficiency and bandwidth density required for the next wave of frontier models.”
From Cadence’s perspective, the collaboration reflects the evolving demands of AI infrastructure at scale. Boyd Phelps, Senior Vice President and General Manager of the Silicon Solutions Group at Cadence, highlighted the role of advanced interconnects in meeting future workloads.
“As AI capacity continues to expand dramatically to accommodate unprecedented demand and workloads, scale-up and scale-out are transforming AI infrastructure,” Phelps said. “Cadence is dedicated to providing next-generation optical interconnect solutions to optimize data center performance and efficiency, and our collaboration with Lightmatter demonstrates our commitment to the evolution of advanced interconnects. By integrating our high-speed SerDes and UCIe IP into this new CPO platform, we’re helping our customers build more scalable, power-efficient AI systems.”
Industry analysts also see the partnership as strategically significant. Roy Chua, founder and principal at AvidThink, pointed to the inevitability of integrated photonic architectures.
“While the industry has long debated the timing of CPO, the move toward integrated 3D-stacked photonic designs is now inevitable,” Chua said. “This collaboration between Cadence and Lightmatter is crucial because it addresses the architectural foundation of electrical-optical connectivity integration. By optimizing advanced CMOS IP with a future-ready CPO design, they are giving the ecosystem the tools needed to escape the limitations of shoreline-bound interconnects.”
Implications for Hyperscalers and Chip Designers
One of the most important aspects of the Lightmatter–Cadence collaboration is its relevance to hyperscalers and semiconductor companies designing custom AI chips.
Hyperscalers such as cloud providers and large technology companies are increasingly building proprietary silicon to optimize performance, cost, and energy efficiency. These custom chips require advanced interconnect technologies to support massive scale-out and scale-up architectures.
By delivering silicon-proven technologies and integration-ready CPO solutions, Lightmatter and Cadence are positioning themselves as key enablers of next-generation AI infrastructure.
The integration of UCIe IP further supports chiplet-based architectures, allowing designers to mix and match compute, memory, and I/O components with greater flexibility. This modular approach is becoming essential as chip complexity increases and monolithic designs become less practical.
Overcoming Technical Challenges in Co-Packaged Optics
Despite its promise, co-packaged optics presents significant technical challenges. Integrating optical and electronic components within the same package requires advances in thermal management, signal integrity, packaging materials, and manufacturing processes.
Optical components are sensitive to temperature and alignment, while electronic components generate heat and require precise electrical connectivity. Balancing these requirements is one of the most complex engineering problems in modern semiconductor design.
Lightmatter and Cadence’s collaboration directly addresses these challenges by combining photonics expertise with advanced CMOS IP and EDA workflows. By leveraging industry-standard packaging techniques and silicon-proven IP, the partnership aims to reduce risk and accelerate adoption.
The Broader Industry Context: From Copper to Photonics
The shift toward optical interconnects reflects a broader transformation in computing. As Moore’s Law slows and transistor scaling becomes increasingly expensive, system-level innovation is becoming the primary driver of performance gains.
In this context, data movement—not compute—has emerged as the dominant bottleneck. Optical technologies offer a path to overcome this bottleneck by delivering orders-of-magnitude improvements in bandwidth and energy efficiency.
Lightmatter’s focus on photonic computing and optical interconnects positions it at the forefront of this transformation. Cadence’s role as a provider of silicon IP and design tools makes it a critical partner in translating photonic innovation into manufacturable products.
The Future of AI Interconnects
The collaboration between Lightmatter and Cadence represents more than a technical partnership—it signals a broader shift in how the semiconductor and AI industries approach system design.
As AI workloads continue to grow, the demand for faster, more efficient, and more scalable interconnects will intensify. Co-packaged optics is increasingly seen as a foundational technology for future AI systems, enabling new architectures that were previously impossible with electrical interconnects alone.
By combining photonics, silicon IP, and advanced packaging, Lightmatter and Cadence are helping to lay the groundwork for this future.
In the coming years, the success of CPO technologies will depend not only on technical breakthroughs but also on ecosystem collaboration. Partnerships like the one between Lightmatter and Cadence demonstrate how companies across the semiconductor value chain can work together to overcome complex challenges and unlock new opportunities.
As AI models push the boundaries of scale and performance, the importance of optical interconnects will only grow. With this collaboration, Lightmatter and Cadence are positioning themselves at the center of the next wave of AI infrastructure innovation—one where light, not copper, becomes the dominant medium for moving data.
Source Link:https://www.businesswire.com/
