Edgecore Networks, 1Finity from the Fujitsu Group, and Liqid have locked in an MOU to work together on a pretty ambitious idea. They want to build solutions that let companies run distributed data centers across huge distances. Not just between racks or buildings but anywhere from a few kilometers out to hundreds of kilometers, all tied together through an all-photonics network that does not get in the way of performance.
The tech they are stitching together leans heavily on software defined composable infrastructure. Basically it lets resources like GPUs, FPGAs, memory, and storage move around and scale on demand even when the physical hardware is spread across multiple sites. You get wide area pooling without the usual friction and you get a lot more freedom to size, shuffle, or optimize resources in real time.
The high bandwidth optical interconnects give them the breathing room to move data fast across long distances. That means enterprises can treat multiple AI data centers almost like one giant cluster. You can expand across racks, across a campus, or across sites without worrying about bottlenecks. It opens the door for AI centers to cooperate instead of sitting in isolated islands and it simplifies expansion from a single rack all the way up to multi campus setups. The whole thing sits on an optical backbone tuned for low latency and heavy throughput which is exactly what AI and HPC workloads demand.
Also Read: Palo Alto Advances Cloud Security with Cortex Cloud 2.0
They are aiming for top tier performance standards too. The design targets high speed and low latency for both RDMA and NVMe over Fabrics without losing throughput, which is usually the first thing to get compromised in long haul setups.
The trio will show the solution at SC 25 in St. Louis from November 16 to 21. If everything stays on track, the offering is expected to hit the market at the end of March 2026.

