How our optical investments support AI

Arelion has long championed optical innovation to improve the scalability and capacity of our global network. These innovations span investments in coherent pluggable optics, IPoDWDM and disaggregated architectures, enabling us to future-proof our network for tomorrow’s connectivity demands.

That time has arrived with the rise of AI, as hyperscalers, enterprises and service providers require a powerful, dynamic network underlay to leverage the power of emerging applications.

Connectivity attributes such as ultra-low latency, high-capacity bandwidth and expansive network reach are non-negotiable for enabling AI’s real-time functionality and high-volume data transport between data centers. Optical innovation has always been integral to our mission, as we were the first Tier-1 operator to offer 100GE Wavelengths on both sides of the Atlantic.

This dedication has remained, as we’ve deployed 400GE in 95% of our European network and 80% of our US network. We were also early adopters of network disaggregation, providing us with multi-vendor architectures and modular scaling to enhance flexibility and expedite network deployments according to our customers’ needs.

These industry firsts and technological benchmarks reinforce the importance of optical innovation in our business model, proving that it persists as a vital component of our DNA. But how do our recent optical networking investments and live network field trials provide the foundation for businesses needing to capitalize on AI’s capabilities?

From the lab to the field: Live network trials

While testing optical equipment in lab settings with “perfect” fiber is one thing, we push ourselves to test these technologies over live fiber in real-world conditions to understand our network’s true capability. In Feburary 2025, we successfully completed a 1,069-kilometer 800-Gbps IP transmission on our long-haul route between Stockholm, Sweden, and Hamburg, Germany, using Cisco 800G ZR+ coherent optics.

This trial demonstrated our ability to lower CAPEX costs by 66 percent and OPEX costs by 95 percent while allowing us to fully utilize the port capacity of Cisco’s next-generation routers. This milestone proved that we’re able to maximize bandwidth while reducing energy consumption, both key factors in enabling power-hungry, high-capacity AI applications.

In December 2024, we achieved a 2,253-kilometer IP transmission with Cisco 400G Ultra Long Haul pluggables along our route from Chicago to Denver, providing longer transmission distances and higher cost savings than currently deployed transponders. These pluggables enable us to reduce CAPEX by 35 percent and OPEX costs by 84 percent when expanding our network, providing wider reach with high capacities that support customers’ AI/ML and cloud applications.

This trial proved that we can build more cost-efficient networks by consolidating excessive hardware and lowering the number of interfaces to maintain, thereby mitigating financial barriers to network expansion.

In August 2024, we also achieved the world’s first live 1.6 Tb/s Wave data transmission with Ciena’s WaveLogic 6 Extreme between our Points-of-Presence (PoPs) in Ashburn, VA, and Telxius’ Virginia Beach cable landing station. This trial increases capacity between two critical technology hubs, enabling us to support the higher traffic volumes and faster connectivity speeds vital to seamless AI operations.

Throughout these trials, we leverage our collaborative partnerships with leading equipment manufacturers to foster further innovation and showcase the true power of these optical technologies in providing massive capacities.

Optical deployments and route enhancements

Aside from trials, we also actively deploy various optical solutions in our live network to enhance services for our customers. In August 2023, we were the first Tier-1 provider to deploy Cisco 400G QSFP-DD Bright Optical Modules across 675 kilometers between Stockholm and Copenhagen over a third-party Optical Open Line System.

This achievement accelerates our mission to provide customers with cost-optimized performance and capacity while ensuring energy efficiency, particularly as connectivity requirements heighten with the proliferation of AI applications.

In June 2023, we doubled the capacity of one of our highest-traffic routes between the technology hubs of Ashburn, VA, and Atlanta, GA. Our upgrade of an existing Infinera FlexILS flexible-grid line system enables us to support L-band transmission and eliminate vendor lock-in, marking our first route with active L-band capacity in service.

This upgrade was integrated to support substantial demand from our customers for terabit-scale bandwidth between these booming markets. Above all, this innovation allows us to optimize the cost per bit and minimize operational expenditures, allowing us to pass the cost savings on to our customers.

Enabling AI on a global scale

Our next blog will explore how our global networking footprint supports AI use cases, empowering our customers with reliable, low-latency connectivity that helps them capitalize on AI’s opportunities no matter where they’re located.

Dariusz Solowiej, VP Network Technology & Customer Operations