Aria Networks has appointed Jonathan Davidson, former Executive Vice President and General Manager of Cisco Networking, to its Board of Directors. Davidson, currently Founder and CEO of Trace Computing, brings more than 30 years of experience in networking engineering, product development and business leadership.

Davidson previously led Cisco’s networking business across enterprise, data center, service provider and cloud markets, including optics, hardware and software. He also headed Cisco’s Mass-Scale Infrastructure Group and Service Provider Networking business. Before returning to Cisco in 2017, Davidson served as Executive Vice President and General Manager of Engineering and Product at Juniper Networks.

Founded in 2025 by Mansour Karam, Aria Networks develops Deep Networking, combining AI-Class hardware, AriaOS, end-to-end telemetry, intelligent agents and cloud-delivered updates to improve token efficiency in AI infrastructure.

“AI factories are the largest buildout of compute in history, and connectivity has become the bottleneck,” Davidson said. He emphasized GPU utilization and tokens per second as measures of infrastructure efficiency.

Analysis

Davidson’s appointment adds experience in large-scale networking products and engineering organizations to Aria’s board. The company aims to correlate network behavior with AI workload performance rather than evaluating connectivity solely through bandwidth and port speeds. The appointment itself does not announce a new product, customer deployment or commercial agreement.

Aria Networks is positioning its Deep Networking architecture as an alternative to conventional data center networking approaches that primarily emphasize switching capacity, port density, and transport performance. Its strategy combines high-bandwidth Ethernet switching hardware with AriaOS, continuous network telemetry, AI-driven analytics, and intelligent agents operating across the network fabric. Rather than treating switches as independent forwarding devices, Aria aims to create an integrated system capable of correlating network conditions with AI workload behavior. The objective is to identify congestion, performance anomalies, and inefficient resource utilization before they significantly affect distributed training or inference operations. This approach reflects a broader industry shift toward increasingly automated network operations as AI clusters grow in size and complexity.

Aria’s business proposition centers on improving the economics of AI infrastructure by making the network accountable for application-level outcomes, particularly GPU utilization and token throughput. For hyperscalers, neocloud operators, and enterprises deploying large GPU clusters, network congestion and operational inefficiencies can reduce the productive output of expensive computing resources. Aria argues that optimizing the network as an integrated hardware and software system can improve the return on AI infrastructure investments. Its opportunity extends beyond supplying high-speed Ethernet switches to providing a software-driven operational platform that continuously measures and optimizes network performance. The commercial challenge will be demonstrating measurable improvements in workload efficiency and operating costs against established networking suppliers, particularly as those vendors introduce their own AI-focused telemetry, automation, and fabric-management capabilities.