en.Wedoany.com Reported - Swedish backbone operator Arelion announced the upgrade of its AI Direct product in Stockholm on May 13, 2026, officially launching its 400G Ethernet Virtual Private Line service. Targeting enterprises, emerging cloud providers, hyperscale users, and wholesale customers, the service addresses network bottlenecks in distributed AI infrastructure through high-capacity private connections. The single-port 400Gbps full-rate access capability provides deterministic bandwidth for large-scale model training, inference, and cross-domain data synchronization, with latency metrics reaching carrier-grade assurance levels.
A dynamic priority routing mechanism forms the core resilience layer of this service, leveraging the AS1299 backbone network to perceive network events in real time and trigger automatic failover to backup paths without any manual intervention. On the service access side, a multi-service single-port architecture is adopted, converging EVPL, cloud connectivity, and IP transit onto the same physical interface. Customers can directly connect to AI data centers, GPU clusters, and hybrid cloud environments, reducing equipment stacking and shortening service provisioning cycles. Arelion's Senior Vice President of Network Operations confirmed in a statement that enterprise AI workloads are shifting from the training phase to production deployment, making end-to-end security capabilities and predictable performance essential for network connectivity.
Jumbo frame support pushes the Maximum Transmission Unit to 9400 bytes, significantly increasing the payload per frame while simultaneously reducing the number of packet fragments and protocol overhead. This design directly enhances throughput, particularly for petabyte-scale data synchronization and distributed checkpoint writing scenarios in AI training. The network protection mechanisms and priority service level architecture construct isolated transmission channels for proprietary datasets, model iteration versions, and inference request traffic, precisely aligning with the stringent requirements of Agentic AI for real-time decision-making links. Applicable scenarios extend horizontally from AI infrastructure to electronic trading platforms, real-time streaming media distribution, and cloud gaming acceleration, where low jitter characteristics translate into quantifiable business benefits.
Arelion's proprietary fiber network spans over 77,000 kilometers, with more than 350 Points of Presence deployed in nearly 130 countries and regions worldwide, serving over 2,000 customers. Globally, 72% of internet routes are carried via its AS1299 backbone, and 95% of end-users in the U.S. and EU markets are just a single hop away from AS1299. Arelion has continuously expanded its AI network infrastructure in recent years, deploying 1.6Tbps wavelengths and 400G coherent pluggable optics between Oslo, Stockholm, and Copenhagen in 2025 to build a Nordic AI superhighway, and completing the world's first 1.6Tbps wavelength data transmission live network trial in collaboration with Ciena in 2024.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
