Accton/Edgecore Networks, 1Finity (a Fujitsu company), and Liqid have announced a strategic collaboration to deliver a next-generation datacenter solution that enables seamless all-photonic network connectivity across long distances, ranging from several kilometers to hundreds of kilometers. The joint solution is designed to maintain high transmission speeds, ultra-low latency, and zero throughput loss for critical protocols like RDMA and NVMe-oF, supporting the growing demands of AI and high-performance computing (HPC) workloads. The three companies will showcase the solution at the SC 25 Summit in St. Louis, Missouri, from November 16 to 21, with demonstrations at Accton/Edgecore Booth 5311 and Liqid Booth 805. The technology is expected to be available for deployment in the fourth quarter of 2025. The solution enables wide-area resource sharing through software-defined composable infrastructure, allowing datacenters to dynamically scale compute, memory, storage, and accelerators like GPUs and FPGAs across geographically distributed sites. This creates a unified, agile infrastructure capable of responding in real time to shifting workload demands. Each partner brings key capabilities to the collaboration: Accton/Edgecore Networks contributes a broad portfolio of open networking and optical solutions, including high-performance switches and transceivers designed for AI-scale datacenter environments. 1Finity provides advanced optical networking platforms with modular, scalable architecture that supports high-capacity, low-latency interconnects. Their expertise in photonics and network automation ensures reliable performance over long distances. Liqid delivers a PCIe Gen5-based software-defined composable infrastructure platform that allows organizations to instantly provision and reconfigure physical resources—such as GPUs, memory, and storage—into bare-metal server systems. This enables up to 100% utilization of critical hardware, improving efficiency and reducing costs per AI inference or HPC task. Together, the companies are building a foundation for distributed AI clusters that can operate across multiple datacenters with the same performance and responsiveness as a single, unified system. This is essential as enterprises expand AI workloads from individual racks to global-scale deployments. Mingshou Liu, President of Edgecore, emphasized the importance of open, programmable networking: “As AI scales from racks to global clusters, we need networks that can keep pace. Edgecore’s switching fabric powers the speed of AI innovation, and with 1Finity and Liqid, we’re creating a new era of collaborative, multi-datacenter infrastructure.” Matsui Hideki, Senior Vice President and Head of Photonics System Business Unit at 1Finity, added: “The future of AI depends on networks that are as dynamic as the workloads they support. Our optical platforms enable composable resources to be shared across campuses and datacenters with minimal latency, laying the foundation for the next generation of distributed AI.” Edgar Masri, CEO of Liqid, said: “AI infrastructure must evolve as fast as the workloads it runs. By bringing composable technology to GPUs, memory, and storage, we deliver instant resource provisioning—whether in a single rack or across multiple sites. This collaboration with 1Finity and Accton/Edgecore gives enterprises the agility and scale that static infrastructure simply can’t match.” The solution is designed to simplify infrastructure expansion, enhance resource utilization, and support the real-time demands of AI, HPC, and other compute-intensive applications. It represents a major step forward in enabling efficient, scalable, and intelligent datacenter connectivity across vast distances.