Arelion is expanding its North American optical and IP backbone with five new Points of Presence (PoPs) in Reno, Nevada; Fargo, North Dakota; Houston, Texas; Atlanta, Georgia; and Miami, Florida. The expansion extends high-capacity network connectivity into emerging GPU compute markets and established hyperscale data center hubs, supporting neocloud operators, enterprises and wholesale customers deploying distributed AI infrastructure.

The new locations extend Arelion’s existing North American backbone through regional infrastructure partnerships and targeted network construction. Reno has two locations supporting greater network diversity, while Fargo provides access to infrastructure supporting GPU and high-performance computing. Houston and Atlanta extend connectivity in established enterprise and hyperscale markets; Miami supports South Florida compute demand and subsea routes to Latin America and the Caribbean.

Arelion provides IP Transit, 400G Ethernet Virtual Private Line (EVPL), optical wavelength services and Managed Optical Fiber Network (MOFN) solutions. These support the movement of large datasets between GPU campuses, cloud regions and enterprise facilities. Arelion says multiple paths can support terabit-scale connectivity, and cites a Structure Research forecast that the global neocloud and AI infrastructure market could reach $186 billion by 2030; that figure is a market forecast, not an Arelion revenue or deployment target.

“We’re seeing significant customer demand in these regions as AI infrastructure expands beyond established data center hubs, with neoclouds and GPU-focused providers bringing compute capacity into new markets,” said Art Kazmierczak, Director of Strategic Sales and Network Development at Arelion.

Analysis

Arelion’s expansion addresses Scale Across connectivity in AI infrastructure. While Scale Up fabrics interconnect accelerators within compute systems and Scale Out networks connect servers within AI clusters, wide-area networks interconnect geographically distributed GPU campuses, cloud regions and enterprise facilities. These links carry datasets, model artifacts, checkpoints and traffic to remote inference or compute resources.

The five-market announcement follows earlier Arelion backbone expansions in the Gulf Coast, Santa Clara and Oklahoma City, as well as the extension of its AI Direct services with 400G EVPL. The technical scope is wide-area transport, not the tightly synchronized, low-latency Ethernet or InfiniBand fabrics inside GPU training clusters. Actual economics and customer benefit depend on route diversity, provisioning time, optical capacity and access to compute campuses. Arelion has not named specific GPU customer installations, committed bandwidth at the new locations or measured latency improvements.

Company Profile: Arelion