“With Telescent’s automated cross-connect system, we’re on track to enable the one of the largest regional fiber backbones to ever be built by offering automatic direct connectivity between Frederick, Maryland and the Northern Virginia Internet ecosystem. With Telescent’s system, we will be capable of processing cross-connects in minutes instead of hours or days, all without a human being touching it.”

“With Telescent’s patented robotic technology in place, fiber optic connections between network devices can be provisioned, tested, and delivered without requiring an individual to be dispatched - and can be completed in mere minutes. We are excited to provide this network automation capability for our customers in the Pacific Northwest, as it also reduces service delivery intervals, offers real-time implementation capabilities, and is supported by sophisticated monitoring tools that allow us to locate cable faults automatically, should they arise.”

“As society becomes more reliant on deep neural networks (DNNs) based on ever larger datasets, there is an urgent demand for more efficient distributed DNN training systems.  At MIT we used a Telescent robotic patch panel to develop and test TopoOpt, a novel network interconnect method to build DNN training clusters. With this approach, we demonstrated TopoOpt obtains up to 3× faster training
iteration time than a Fat-tree which can have a significant impact in machine learning applications.”