Arelion Adds GPU Connectivity in Five US Markets

The new Points-of-Presence extend high-capacity links to emerging AI compute campuses, with the $186 billion neocloud market projected by 2030.

Updated on Nov. 11, 2026 in Data Centers

Isometric editorial illustration of bundled fiber-optic cables converging from rows of equipment racks onto a single junction tray.
Arelion added five new network Points-of-Presence in Reno, Fargo, Atlanta, Houston and Miami to link emerging AI GPU compute campuses. AI Illustration. Upload story photo >

Arelion, a network provider, has added five new Points-of-Presence (PoPs) in North American regional data center markets: Reno, Fargo, Atlanta, Houston and Miami. The services offered at these sites include IP Transit, 400G EVPL Ethernet, Wavelengths and Managed Optical Fiber Network, and the expansion is an announced deployment already reflected in the company's network.

Why it matters

AI infrastructure is spreading beyond established data center hubs as neoclouds and GPU-focused providers bring compute capacity into new markets. Connectivity has to follow that capacity, and Arelion is extending high-capacity links to these emerging compute campuses.

Arelion's services are IP Transit, 400G EVPL Ethernet, Wavelengths and Managed Optical Fiber Network, with the 400G Ethernet service rated at 400 gigabits per second. The projection of critical IT capacity from about 2.3 GW to 150 GW is a Structure Research estimate, not a measured result.

The players

Arelion

A network provider whose North American backbone supports IP Transit, 400G Ethernet and managed optical fiber services for high-capacity data traffic.

Structure Research

A market research firm that projects the global neocloud and AI infrastructure market will reach $186 billion by 2030.

The details

A Point-of-Presence is a physical site where a network operator interconnects its backbone with customer systems and other networks. Arelion is reaching compute campuses by combining its existing North American backbone, relationships with regional infrastructure providers, and targeted new network construction. The company says this reduces latency and improves redundancy for enterprises moving workloads between on-premises systems, cloud environments and distributed GPU resources.

Timeline

  1. 2030 — Structure Research's projected date for the $186 billion neocloud and AI infrastructure market.

The Tech Race

Structure Research forecasts the neocloud and AI infrastructure market at $186 billion by 2030, with critical IT capacity rising from about 2.3 GW to 150 GW. Arelion's five new PoPs are positioned to carry traffic for that build-out in markets beyond the traditional data center hubs.

Enterprises moving workloads between on-premises systems, cloud environments and distributed GPU resources may see lower latency and better redundancy as the new PoPs come online. The extraction gives no date for when the Reno, Fargo, Atlanta, Houston and Miami sites will be available to customers, so the timeline is unannounced.

The takeaway

Watch whether Arelion announces activation dates and site capacities for the five new PoPs, which would show how quickly connectivity is following GPU capacity into new markets. The Structure Research projection of $186 billion by 2030 is the benchmark to track as those builds progress.

Further reading

For the wider buildout behind this expansion, see Data Centers coverage across the United States.