Arelion’s AS1299: Tier-1 Reach Requires Constant Renewal

Summary

• Arelion is the privately held carrier formerly known as Telia Carrier; Twelve99 persists as a technical and historical name, while AS1299 is the global routing entity.
• AS1299 reaches the entire Internet via customer routes and settlement-free peering, but Tier-1 status does not guarantee shortest paths, superior support, or immunity from failure.
• Beyond IP transit and dedicated Internet access, it offers Ethernet, wavelengths, managed optical networks, cloud/IX connectivity, DDoS mitigation, mobile, voice, and AI Direct.
• Accumulated routes, fibre, PoPs, peers, and operational knowledge create advantage, yet the need for ongoing investment in capacity, security, and expansion imposes a burden without standalone financial disclosure.

Four Names Represent Four Distinct Layers

The business started within the Telia group and for many years operated as Telia International Carrier, later Telia Carrier. Polhem Infra completed the acquisition of Telia Carrier on 1 June 2021, transferring control from Telia Company. The carrier adopted the Arelion name in January 2022. These dates mark corporate-history boundaries. Describing the company as Telia-owned before the completion is correct; describing the current company as under Telia’s umbrella is not.

Twelve99 did not vanish with the rebrand. The name lives on in the technical domain twelve99.net, which includes a public looking glass, and remains tied to AS1299. This continuity is useful for engineers because autonomous system numbers, hostnames, route filters, customer configurations, and operational references often need stability that outlasts corporate brands. It does not, however, create a second company. Twelve99 is technical continuity; Arelion is the current commercial and corporate identity.

AS1299 is yet another layer—the autonomous system identifier through which the backbone originates, receives, selects, and advertises routes. An ASN is neither a legal entity nor a physical cable. It identifies a routing domain whose policies are expressed across routers, PoPs, and interconnections. Arelion operates that domain, but the underlying paths can ride on fibre owned or contracted under different arrangements.

Distinguishing these four layers—old brand, current company, technical hostname, and routing system—prevents common errors. It stops historical continuity from being turned into a false ownership claim, stops technical identifiers from being treated as subsidiaries, and keeps network-scale figures attributed to the company that publishes them. It also reveals a central feature of Arelion’s history: corporate control changed faster than the backbone’s operational identity.

Tier-1 Network Is a Relationship, Not a Decoration

The phrase “Tier-1 backbone” compresses several facts into a single label. At the routing layer, a Tier-1 autonomous system can reach the whole Internet through routes learned from customers plus settlement-free peering with other large networks. It does not need to buy upstream transit to obtain normal global reachability. That status matters because it removes one supplier dependency and allows the network to sell full-route transit to others. But it is not a title awarded once and kept forever; it is a set of relationships that must be maintained.

Peers change policies, traffic shifts, and carriers must keep enough capacity and direct relationships for settlement-free exchanges to remain mutually rational.

This definition does not certify latency, support quality, DDoS performance, enterprise access, or financial strength. BGP chooses paths based on policy and attributes, not on a universal notion of shortest distance. Even a Tier-1 network depends on colocation providers, submarine cables, access carriers, router vendors, optical equipment suppliers, and the customer traffic that makes the economics work. The Tier-1 label signals a particular routing independence under a specific set of interconnection relationships. It is a starting point for analysis, not a conclusion.

Arelion’s current shape makes this distinction especially visible. The company’s commercial claim is not just that AS1299 has global reach; it is that it can turn that reach into IP transit, private packet transport, optical services, cloud/IX connectivity, network-layer security, and operational support. Hence a customer experiences Tier-1 not as a badge but through a service contract whose value depends on the boundary point, region, capacity, routing policy, and failover capability.

From National Telecom Lineage to Global Carrier

Arelion describes its backbone as having developed organically from 1993. The network grew within the Telia lineage, but its operational logic was international. A national telecom serves retail, mobile, and enterprise customers in a single-country market. A global carrier must stitch together other networks across borders, place routers in neutral facilities, secure long-distance paths, maintain optical capacity, and negotiate interconnections with organisations that are sometimes also competitors.

This difference is what made Telia Carrier a business that could be separated from its former parent. By 2020 it had customer and infrastructure relationships that extended well beyond Telia’s domestic business. Telia Company agreed to sell to Polhem Infra in October 2020, and the acquisition closed in June the following year. It is the completion, not the announcement, that marks the transfer of control. Polhem Infra described the deal as preserving a strategic network relationship with Telia after separation, showing that corporate independence did not erase commercial interdependence.

The 2022 rebrand gave the standalone carrier a name with no tie to the old parent. But the underlying asset was not a newly assembled network; it was a mature backbone that had accumulated fibre paths, PoPs, customer connections, and peering relationships over decades. That matters because connectivity cannot be reproduced merely by introducing a software control plane. New entrants can lease capacity in some markets and light up ports in short order, but they cannot instantly create the direct paths, operational trust, and failure-recovery experience embedded in a globally run long-established network.

Independence also changed the investment question. Under Telia, the carrier competed for capital inside a diversified telecom group. Under Polhem Infra it became a focused infrastructure investment asset backed by Swedish pension capital. That structure may suit long-duration investment, but it does not remove return targets or make capital unlimited. What changes is who decides, how the asset is positioned, and what the outside world can see.

The Backbone Is Not a Single Asset but a Chain of Control

Arelion reports more than 80,000 kilometres of fibre, over 350 PoPs, and service in 129 countries. It speaks of more than 2,000 customers and about 450 access partners. Because each number represents a different layer, they must be treated as company claims. Fibre length indicates physical reach; PoPs indicate interconnection and service-delivery points; country count indicates commercial availability; access partners indicate the ability to deliver beyond direct location. Adding the numbers together yields no meaningful composite measure.

The physical chain can include fibre Arelion owns, fibre it controls through long-term agreements, fibre it leases as capacity, the optical transport systems riding on top, the routers and switches, the colocation space and power, the cross-connects to customers and peers, the submarine-system capacity, the cloud on-ramps, and the local tails that other carriers supply. Arelion may control the service and the routing policy while sharing suppliers and physical failure domains with others.

A “global backbone” is an operational system that combines multiple forms of control; it is not evidence that a single company owns every duct, cable, and building.

This is an operational distinction, not a verbal one. Two logical paths that appear diverse on a network map can share the same duct or submarine cable. Two services sold separately can ride the same router, line system, and facility power. Conversely, leased infrastructure can be highly resilient if the contracts, monitoring, and physical diversity are right. Ownership alone cannot certify reliability; what matters are the failure boundaries and the operator’s ability to observe and recover.

Thus Arelion’s commercial foundation starts from the chain it controls. One must understand which segments are its own backbone, where it depends on facilities, where it relies on access partners, which handoffs are controlled by cloud providers, and where contractual SLAs change. What a customer sees as a single order and support relationship can involve multiple organisations during a fault. Part of the carrier’s value lies in making that chain manageable without pretending it is a single thing.

Fibre Creates Routes; Optical Technology Creates Usable Capacity

Fibre length is a geographic measure, not a capacity figure. One strand can carry many wavelengths, and each wavelength can be upgraded with newer coherent optics and line systems. The same physical path can carry significantly more traffic after an equipment refresh without digging up the route. Arelion’s roll‑out of 400G and higher-capacity upgrades belongs to this optical layer. Transponders, coherent pluggables, amplifiers, and spectrum design turn glass into a working transport path.

Wavelength services provide a dedicated optical channel over a qualified route, suited to predictable, high-volume data movement such as data-centre interconnect, replication, and carrier aggregation. Customers get a clearer capacity boundary than with public IP transit, but the wavelength still depends on the underlying physical fibre, optical hardware, and path protection. “Dedicated” refers to the channel, not to a customer-owned physically isolated cable.

Arelion also sells optical fibre and Managed Optical Fibre Network services. With managed services, the carrier takes on more of the design, equipment, and operational responsibility. Large enterprises, cloud providers, and service providers can avoid building their own optical engineering organisation for every route. In return they become more dependent on Arelion’s geography, supplier choices, restoration procedures, and lifecycle decisions.

Optical services also illustrate why an older backbone cannot be treated as a finished asset. Traffic grows, interfaces get faster, and customers demand new paths to clouds and data centres. Capital must be deployed ahead of each upgrade, before demand is certain. Under-investment leads to congestion and quality erosion; over-investment leaves optical hardware, ports, and long-term contracts underutilised. The carrier’s financial discipline is embedded in capacity planning long before customers notice a change in speed.

PoPs Turn Long-Haul Capacity into a Market

A PoP is the place where backbone capacity becomes available for interconnection. Inside a data centre or carrier hotel, a PoP can contain routers, optical equipment, switches, customer ports, and cross-connects. Dense PoPs shorten the distance between Arelion and its customers, peers, clouds, and IXs. At the same time they increase operational dependency on facility power, cooling, building access, and cross-connect processes.

The “350‑plus” PoP count is a scale claim; it does not mean every location has every product and every speed. A site might offer IP transit but no specific wavelength path, 400 Gb/s Ethernet might need an engineering review, cloud connectivity depends on a provider’s on-ramp, and a remote enterprise location requires a partner tail. The procurement question is not “Is Arelion in this country?” but “Which services are available at this handoff point?”

PoP density also affects connectivity. Direct interconnection reduces the number of intermediate networks a path must cross. More direct customers and peers increase route choice. Arelion states that 95% of end users in North America and Europe are reachable within one network hop, and that AS1299 is the world’s most connected backbone on selected metrics. Because results vary with dataset, definition, and date, these must be treated as company claims. One‑hop reach does not mean lowest latency for every flow, and connectivity is not traffic market share.

The Packet Layer Carries Multiple Products Over Shared Infrastructure

Above the optical layer, routers and packet systems turn capacity into routable and private services. AS1299’s BGP control plane exchanges Internet routes. MPLS and segment routing underpin traffic engineering and private packet transport. Ethernet provides Layer‑2 connectivity, Smart IP-VPN offers a managed routed WAN, and Cloud Connect and IX Connect extend those paths into specific ecosystems. Each product shares some of the underlying base while presenting different service boundaries to the customer.

Arelion’s Ethernet includes point‑to‑point EVPL and multipoint ELAN from 10 Mb/s to 400 Gb/s, subject to path and technical availability. Its materials describe using segment routing and Flex‑Algo for select low‑latency paths within the MPLS backbone. These mechanisms can increase predictability inside the carrier domain but do not eliminate access tails or external handoffs that Arelion does not fully control.

Service‑level figures make this boundary visible. Arelion advertises 99.999% availability for MPLS backbone services, 99.99% for basic inter‑PoP Ethernet, and 99.5% end‑to‑end when managed NIDs and access tails are included. The numbers are not interchangeable. The lower promise for the broader scope reflects the additional hardware and third‑party dependencies. Actual terms vary by contract and route.

Customers can also use logical separation to run multiple services over a single port, which reduces cross‑connect and interface cost while concentrating them onto one physical failure domain. A port failure can silence several logically separate lines at once. Efficient multiplexing and concentrated risk deserve to be described together.

IP Transit Is the Premier Commercial Expression of Tier-1 Status

IP transit customers usually operate their own autonomous system, advertise prefixes via BGP, and receive Internet routes from Arelion. The carrier carries outbound traffic to other networks and delivers inbound traffic back to the customer prefixes. The product is priced and contracted as reachability, capacity, and service, but its foundational value lies in the customer and peer relationships of AS1299.

Because BGP hides physical paths behind route advertisements, the transaction can look simple. In reality Arelion must maintain a full routing table, filter invalid and unauthorised announcements, balance traffic across links, provision capacity, protect sessions, manage communities, and recover from failure across dozens of PoPs. A customer that multihomes to another carrier gains redundancy and path choice but also inherits more complex routing problems: preference, inbound‑traffic control, and understanding how failure propagates at each carrier.

Tier-1 economics does not mean all interconnections are free. Settlement‑free peering removes payment for specific exchanges with qualifying networks. Arelion still spends on fibre, facilities, hardware, power, access, staff, maintenance, and other commercial relationships. It also buys local services and partner access when that is more efficient than building directly. The absence of an upstream transit bill is one cost‑structure factor, not a zero‑cost network.

Route position must also be defended continuously. If traffic balance, geographic reach, or commercial policy changes significantly, peers may seek to adjust terms or end the relationship. The practical barrier to entry is thus accumulated and continuously guarded. What keeps other large backbones in a settlement‑free exchange is enough customers, direct reach, capacity, and operational credibility.

Dedicated Internet Access Shifts More Routing Responsibility to the Carrier

Dedicated Internet access (DIA) serves enterprises that want a managed connection without running a full BGP relationship and global routing table themselves. Customers receive a default route or a managed edge, and Arelion takes on more of the Internet‑connectivity design. At some locations the physical access and quality may resemble transit, but the operational contract is different.

The distinction matters during failure and change. A transit customer can use its own ASN, prefixes, communities, and multihomed policy. A DIA customer typically has a narrower control surface and relies more on the carrier’s routing and edge support. Neither is inherently superior; transit suits organisations that need routing capability and policy control, DIA suits buyers who want a managed boundary.

SecureConnect bundles Internet access or transit with automatic DDoS protection. For customers who would otherwise contract transport and mitigation separately, it reduces procurement and configuration boundaries. The protection scope, however, remains bounded; it does not protect credentials, endpoints, application logic, or attacks outside the chosen network‑layer service.

BGP Communities Make Routing Policy Part of the Product

BGP communities are tags attached to route advertisements. Arelion interprets customer tags and can apply carrier‑defined actions such as preference changes, limiting propagation scope, AS‑path prepending, or activating blackholing. Customers can influence the network without direct access to the carrier’s routers.

This is a powerful delegated control. An operator can make one prefix less preferred in a region, restrict where it is advertised, or sacrifice reachability to protect the wider network during an attack. The meanings are specific to AS1299; the same community values at another carrier may signal different behaviours. Customers need up‑to‑date documentation, controlled change, and tools to verify the result in a routing table.

Communities themselves carry failure risk. A misapplied tag can drop reachability or steer traffic onto unintended paths. A remotely triggered blackhole intentionally discards traffic toward a target upstream so that an attack does not saturate shared links. It is effective in an emergency because it trades one target’s availability for network stability, but it is not free mitigation.

A Looking Glass Shows a Slice of Routing Reality

Arelion’s Twelve99 looking glass allows route, ping, and traceroute checks from selected network locations. It answers practical questions: how does AS1299 see a given prefix, which path does it select from a particular vantage point, and where does latency appear from that perspective. The company kept the Twelve99 technical identifier through the rebrand precisely for this operational continuity.

A looking glass is evidence from a single observation point, not a network‑wide audit. It does not show every alternate path, past failure, internal preference, or physical fibre route. BGP policy can differ by location, and traceroute can hide hops or respond differently under load. It is most useful when combined with customer‑side measurements, route collectors, service tickets, and carrier performance data.

Arelion also publishes monthly IP‑network performance metrics. Buyers can track operator‑chosen measures over time, though the methodology and scope are defined by the company. Publishing operational data raises transparency, but independent assurance still requires clear definitions and external observation.

RPKI Does Not Fix BGP Entirely; It Reduces One Routing Risk

Route Origin Authorisation lets a prefix holder state which autonomous system is authorised to originate that route. Route origin validation can classify announcements as valid, invalid, or not found. Filtering invalid origins can reduce some hijacks and misconfigurations. Arelion’s materials position RPKI as one element of routing security.

The mechanism checks origin authority for the prefix in question. It does not validate the full AS path, does not guarantee that peers export routes correctly, and does not prevent every route leak. A valid origin can still propagate through an unintended path, and operational errors can still occur in filters, route objects, and customer configuration. RPKI narrows a trust problem but does not replace BGP policy, monitoring, or incident response.

For a Tier-1 carrier the boundary matters because routing security and reachability can conflict. When records are wrong, an overly strict filter can block legitimate traffic; an overly loose filter can propagate invalid announcements. Staged policy, customer outreach, exceptions, and current data are required. Phrases like “RPKI makes BGP safe” hide the operational judgement that remains.

IX Connect Sells Access to Interconnection, Not Reach to the Whole Internet

Internet exchanges provide a venue where networks establish peering. Arelion’s IX Connect transports a customer from an available Arelion location to an IX port, enabling remote participation without building a separate local presence. For networks entering a new region, it can reduce the cost and time needed to reach multiple peers.

The transport service itself does not create the peering policy. Customers still must meet IX participation requirements, provision a compatible port, establish bilateral or route‑server arrangements, and maintain their own route filters. A remote path may also add one more dependency compared with physically placing a router at the IX. IX Connect is therefore an access service to the routing marketplace, not a substitute for a customer’s own interconnection strategy.

This product also illustrates why a Tier-1 carrier both competes and cooperates with IXs. Arelion peers at exchanges while selling transport to them. A customer can replace some paid transit with direct peering while still buying Arelion capacity to reach the IX and the rest of the Internet. The boundaries among transit, peering, and transport are commercial choices built on the same physical network.

Cloud Connect Carries the Backbone to the Boundary Controlled by the Cloud Provider

Arelion lists private connectivity to AWS, Microsoft Azure, Google Cloud, Oracle, and IBM. Cloud Connect carries customer traffic to an eligible cloud on-ramp, reducing reliance on public Internet paths for that segment. Customers can gain more predictable path, capacity, and security than sending all cloud traffic over general transit.

The service terminates at a shared boundary. The cloud provider controls the virtual interface, regional availability, quotas, and internal fabric. The customer controls accounts, routing, and workloads. Arelion controls the carrier path it sells. Failure can occur on either side, so configurations must be aligned. Calling the whole path “private” can create an illusion of single‑ownership where there is none.

Cloud connectivity also shifts the carrier’s competitive landscape. Hyperscalers operate large private backbones and carry more traffic internally among their own regions. Arelion’s opportunity lies in connecting enterprises, data centres, multiple clouds, and networks that do not share a single administrative domain. That is also its ceiling: it cannot dictate what happens inside the cloud and cannot replace the cloud provider’s own fabric.

DDoS Defence Has Become Part of Backbone Capacity Planning

High‑volume attacks consume link capacity, router resources, and scrubbing capacity. A global carrier can observe enough traffic to divert attack flows before customer access links become congested. Arelion’s DDoS offering combines detection, diversion, and scrubbing, returning permitted traffic to its destination.

In a report published on 15 July 2026, Arelion stated that the Aisuru botnet accounted for roughly one‑third of attack traffic observed in its dataset, with the largest attack reaching 6.1 Tb/s. This is an operator observation based on Arelion’s network and methodology. It is useful because it shows the attack scale visible from a large backbone, but it is not a census of global DDoS activity.

The 6.1 Tb/s observation should not be converted into a guaranteed mitigation capacity. A carrier may observe an attack across the whole backbone without any single customer receiving that full volume, and service commitments vary by configuration and contract. Effective mitigation also depends on detection time, BGP convergence, scrubbing placement, clean‑path capacity, and false‑positive control. Application‑layer attacks can slip past volumetric filters because individual packets look legitimate.

Remotely triggered blackholing is a design‑time emergency tool. A community can cause traffic toward a targeted prefix to be discarded upstream. It protects shared capacity while making the target unreachable. Scrubbing aims to keep the service up; blackholing accepts an outage to contain damage. A mature DDoS product needs both, clear activation conditions, and customer authority over when the nuclear option is used.

SecureConnect bakes mitigation into the connectivity purchase instead of adding it as a separate system after the line is bought. That makes it easier to have protection enabled before an incident. Strategically, what to watch is not the bundle name but attachment rates, demonstrated mitigation, scope of protected services, and whether the architecture can scale as peak attacks rise.

AI Direct Bundles Wide-Area Transport for Distributed AI Systems into a Single Proposition

AI Direct is Arelion’s connectivity portfolio for moving data among AI clusters, data centres, and clouds. It bundles Ethernet, wavelengths, Internet access, managed optical networking, and security—services the carrier already offers—into an AI‑infrastructure proposition. It does not supply GPUs, storage, or model‑training software. Its role begins where data leaves a single facility or administrative domain.

This boundary has an economic reason. Training inside one campus relies on local high‑performance fabrics whose latency and collective‑communication characteristics a wide‑area carrier cannot replace. Wide‑area demand arises when datasets, checkpoints, replication, inference traffic, and whole workloads move between facilities. Arelion’s strength is reach among locations, so it must connect to internal AI fabrics rather than become one.

The company added 400G EVPL to AI Direct in May 2026. The announcement demonstrates a high‑capacity packet option on qualifying routes, though it does not mean 400G is available at every location. It also announced Danish network upgrades and cable‑landing‑station environment enhancements in April, positioning them as part of a Nordic AI corridor. Earlier expansions included a North American PoP in Oklahoma City and, in 2026, extended sales reach through a Mexican channel programme. All these show active route‑and‑product strategy, but the carrier does not disclose per‑case utilisation, customer concentration, or investment amounts.

The strongest evidence that AI Direct is more than a label would be named high‑capacity customers, lit ports, sustained traffic, and route‑specific service commitments. Announcing an “AI superhighway” communicates strategic intent but does not show how much revenue, utilisation, or new capacity AI workloads have generated. The current evidence supports a credible transport portfolio and investment direction; it does not yet prove that AI has already transformed the company’s economics.

Enterprise Services Monetise the Backbone Beyond Transit

Transit prices face long‑term downward pressure as capacity improves and alternatives grow. Arelion can respond with services that add more control, boundaries, and support onto the same paths. Ethernet, Smart IP-VPN, Cloud Connect, IX Connect, managed optical networking, and DDoS protection let it solve a wider set of problems for a single customer.

A broad portfolio does not mean every service has the same margin or the same customer. Wholesale carriers may buy transit and wavelengths, multinationals may buy managed access and IP-VPN, cloud providers may buy optical capacity, and regional ISPs may buy transit and IX access. A single sales organisation must sell common infrastructure through different contracts and support models.

Breadth also complicates internal operations. Capacity reserved for private services, public transit, and mitigation must be planned together. An optical‑path change affects multiple packet products, and a local‑tail delay can block an otherwise automated global order. Asset reuse improves economics only if service separation, change management, and capacity accounting are precise.

Arelion’s mobile data, IoT, voice, and messaging offerings broaden the portfolio further. The research evidence is strongest on backbone and connectivity products, so these adjacent services should be treated as part of the current catalogue rather than as a basis for unverified market‑share claims. The strategic implication is that Arelion is not a pure IP‑transit wholesaler even though AS1299 remains the technical centrepiece.

Access Partners Extend Reach but Also Change SLA Boundaries

Arelion reports about 450 access partners. When direct backbone fibre does not reach a customer building, those carriers link the building to an Arelion PoP. This model extends service coverage without building every last mile. Customers get one commercial relationship for a path that is assembled from multiple carriers.

Access tails are the least standardised part of the service. Delivery times, repair procedures, available bandwidth, jumbo‑frame support, demarcation equipment, and local regulation differ. Even when Arelion monitors and manages the tail, the physical repair may be performed by another carrier. The lower published availability for end‑to‑end Ethernet versus inter‑PoP reflects this extra dependency.

Worldwide service coverage should therefore be read as a capability to deliver, not as a map of owned facilities. Enterprise buyers need to identify local suppliers, verify path diversity, understand escalation rights, and check whether a second line shares the same duct. A single global contract can simplify governance but does not change the local physical conditions.

Operations Turn Network Assets into Services Customers Can Use

A backbone has value only if it can be provisioned, observed, and repaired. Arelion’s customer model includes the MyArelion portal, service management, and network operations support. These mechanisms translate ports, paths, tickets, planned maintenance, and performance into customer experience.

Automation can speed delivery and make state visible, but a global carrier cannot reduce every event to an API call. Fibre restoration needs field crews, cloud handoffs can reject configuration, peers change policy, colocation facilities require cross-connects, and regional carriers miss delivery dates. The quality of carrier automation is determined not only by how quickly it takes an order but by how clearly it surfaces exceptions and accountability.

Arelion reports customer‑experience awards and a high NPS. That can signal an intentional service culture, but methodology and respondent population are not independently audited in the provided materials. Awards and surveys should sit alongside operational detail, not replace it.

The Business Model Reuses a Single Infrastructure Base Across Multiple Boundaries

Arelion earns revenue by selling access to capacity, paths, service guarantees, and operational responsibility. IP transit monetises global routing reach; Ethernet and IP-VPN monetise private packet paths; wavelengths and managed optical networking monetise optical capacity and engineering; cloud and IX products monetise ecosystem access; DDoS services monetise visibility and protection. Multiple products share the same backbone but each starts from a different boundary.

This reuse can raise asset utilisation. A single fibre path and PoP can support more revenue the more products and customer groups it serves. But separation and headroom are needed. Planning every service at the same optimistic utilisation rate can reveal hidden contention when an attack or traffic shift arrives. Capacity accounting must distinguish reserved ports, ordinary traffic, protected traffic, and failure scenarios.

Commercial leverage comes from reducing the customer’s integration work. A multinational can separately assemble local carriers, IX ports, cloud lines, transit, optical paths, and security. Arelion tries to integrate more of that chain. Customers pay not only for bits but for fewer contracts, a clearer support path, and transferred responsibility. The carrier captures a premium only when the integrated service works better than the disaggregated alternative.

Pension Capital Ownership Changes the Time Horizon but Not Economic Laws

Polhem Infra owns Arelion. Current descriptions say Polhem Infra is jointly owned by Sweden’s Third and Fourth AP Funds. The carrier is thus tied to national pension capital through an infrastructure investment vehicle. This structure suits an asset whose route and interconnection position is built over long cycles.

Long‑term capital can more comfortably fund investments that take years to recover, such as fibre rights, optical refreshes, and new PoPs. It may also prize steady cash generation more than short‑term equity‑market messaging. But the owners still have fiduciary duties and return requirements. The pension linkage is not a guarantee that routes will be funded regardless of demand, nor does it reveal how much of the capital backing Arelion is debt, operating cash, or owner equity.

Since 2021, governance changes have been more visible than technical changes. AS1299 has kept routing while board control, funding, and strategic priorities moved from Telia to Polhem Infra. Daniel Kurgan became CEO in October 2023, and Charles Gill joined as CFO in March 2024. The current management page shows a broad team spanning strategy, sales, legal, people, technology, and operations, though it is not a full statutory board list.

Not every design choice can be credited to individual executives. The backbone is the cumulative output of decades of engineering, operations, procurement, sales, and partner teams. Management decides capital allocation, risk appetite, product direction, and disclosure. Day‑to‑day quality depends on distributed expertise and institutional memory.

Private Ownership Leaves Key Financial Questions Unanswered

From the evidence supplied one cannot determine Arelion’s standalone current revenue, profit, debt, capex, valuation, traffic volume, contract tenor, or customer concentration. The gap matters because a global carrier must continuously fund equipment, rights‑of‑way, colocation, access, and security capacity. Product announcements show activity but not investment returns.

Past deal information describes a 2020–2021 sale, not the 2026 enterprise value. Keeping an old transaction figure at face value ignores changes in traffic, assets, capital structure, and market conditions. Equally, connectivity rankings cannot serve as a proxy for revenue or margin. A network can be highly connected while operating in a market where per‑bit prices are falling.

This opacity changes how performance must be judged. Public routing data and the looking glass can show reachability and path behaviour. Press releases show route and product activity; customer announcements show some adoption. But they do not show whether capital is being allocated at adequate returns, whether a single customer dominates revenue, or whether leverage constrains the next refresh.

The conclusions one can draw are bounded. Arelion looks like an active global carrier with a broad product range and continued investment. The supplied materials do not support margin estimates or the claim that pension ownership guarantees the next capital cycle.

The Economics of Refresh Lies Between Physical Ownership and Contractual Control

Investment decisions are more complex than a binary choice between owning and leasing a particular route. A Tier-1 carrier must maintain multiple modes of control simultaneously. Fibre ownership gives direct authority over the physical path but still depends on permits, power, restoration crews, landing stations, and hardware suppliers. Long‑term fibre rights can look like ownership for planning purposes while locking in contractual obligations for years. Short‑term capacity deals preserve flexibility but expose the carrier to repricing, supplier availability, and renegotiation.

Which form is appropriate depends on the route, the market, and the consequences of failure.

This matters because network economics is shaped before services are sold. Arelion may need to pre‑place line systems, router capacity, cross‑connects, and headroom before new customers commit traffic. Expanding into a new PoP requires facility contracts and hardware ordering ahead of demand, even when demand depends on a forward pipeline. A high‑capacity AI corridor may be compelling at the market level, yet individual customers may not commit to where workloads will land until years later. Capital is committed before the revenue that will prove the case.

Resilience has the same timing gap. Protection paths are most valuable when the primary path fails, so a portion of capacity appears lightly used during normal operation. What looks like idle headroom to the finance function looks like protection to operations. The network must judge how much headroom is economically justified against the customer contracts and failure scenarios it is meant to protect. Too little, and a fibre cut or hardware failure leads to customer loss. Too much, and returns are diluted unless that reliability can be reflected in price.

Arelion’s broad portfolio complicates the calculation further because the same physical and packet substrate supports products with different load characteristics. Transit is bursty and price‑sensitive; Ethernet and wavelengths may carry large contracted flows. DDoS mitigation needs capacity that is rarely used until an attack arrives. Cloud connectivity depends on the location and growth of external platforms. Capacity planning cannot be a simple sum of average traffic; it must account for correlated peaks, failover shifts, maintenance, and the possibility that multiple services migrate onto the same survivable path.

The 2026 AI Direct positioning adds one more uncertainty. To the extent that inter‑facility data movement generates real WAN demand, Arelion has a reason to tie high‑capacity Ethernet and optical services to distributed AI infrastructure. Yet even AI demand does not exempt a carrier from normal investment discipline. Paths must connect real facilities; ports must be lit; customers must pay for capacity; and traffic must persist long enough to justify the hardware and contract rights. A fashionable workload does not change the economics of unused capacity.

Supplier strategy also shapes the refresh cycle. Routers and optical platforms are not infinitely interchangeable. A large refresh can lock in operational familiarity, sparing plans, software dependencies, and manufacturer support for years. Open line systems and coherent pluggables can lower lock‑in by decoupling parts of the optical stack, but interoperability still requires design and testing. Choice is valuable only if the organisation maintains the skills and operational procedures to exercise it.

Facilities carry a similar stickiness. A PoP gains value the more customers, peers, and clouds connect there, but that very density makes it hard to move. Cross‑connects, customer gear, maintenance routines, and commercial relationships accumulate in one location. Even when Arelion owns the routers inside, the data‑centre operator can have negotiating power. The question is not just whether a carrier has an alternative site but whether it has enough options, route diversity, and commercial leverage to avoid one building or one operator becoming a hidden strategic constraint.

This is why the financial opacity of private ownership matters for infrastructure analysis. Without current capex, debt, lease obligations, and cash flows, an outsider cannot tell how much of the refresh is being funded from operating cash, owner capital, or borrowing. Nor can one judge how much headroom exists before the next hardware generation must be funded. Technical announcements demonstrate that investment is being made; they do not prove that the same pace can be maintained if the pricing environment worsens, demand slackens, or resilience requirements rise.

The strong external evidence is cumulative: new routes that are actually completed, not just announced; new PoPs that come with customer demand; current‑generation interfaces that become orderable; direct paths that remain stable; and a security substrate that handles larger attack peaks. None of these alone provides a margin, but together they show whether capital is being turned into an operational system that keeps AS1299 economically relevant.

For Arelion, the refresh obligation and the barrier to entry are inseparable. Part of the advantage is a network position accumulated over more than 30 years, which saves customers from having to assemble the same reach themselves. But that long history also creates a large installed base that must be kept current. The more AS1299 gains value as a neutral global path, the more costly it becomes to defer investment. Tier-1 status therefore creates not just bargaining power but an ongoing capital obligation.

Geography Is a Map of Services, Routes, and Dependencies

Arelion’s origin and ownership base are Nordic, and its backbone extends across Europe, North America, and into Asia. The company reports service in 129 countries. This is a commercial reach figure, not a count of owned fibre systems or direct PoPs. Buyers need three maps: where Arelion has backbone and facilities, where it can deliver via partners, and where specific services are technically available.

Europe has high density in the network’s lineage. North America is a major transit and enterprise market, where recent additions such as Oklahoma City were tied to cloud and AI data‑centre demand. The 2026 Denmark investments concerned routes and cable‑landing‑station environment important for Nordic and submarine connectivity. In Mexico, route development and a channel programme widened access to a growing market. The evidence pack provides less detailed route information for Asia than for Europe and North America.

Submarine connectivity needs separate handling. Arelion can buy spectrum or capacity, place equipment in a landing station, and run end‑to‑end services without owning an entire submarine system. Cable faults, landing‑station incidents, and shared terrestrial backhaul affect multiple carriers. Logical path diversity must be verified against cable and duct diversity.

Geopolitics also shifts route economics. Cross‑border regulation, sanctions, licensing, cable security, and national resilience policies affect where a carrier can build and how customers perceive risk. The evidence pack does not support specific current disputes involving Arelion, so these should be treated as structural constraints, not allegations.

Competition Occurs Simultaneously on Multiple Layers

In the Tier-1 global transit layer Arelion competes with NTT, Lumen’s Level 3 lineage, GTT, Tata Communications, Cogent, Sparkle, and others. Direct comparison requires dated evidence on peering relationships, geography, latency, capacity, security, support, and price. No single connectivity ranking determines choice.

In the physical and optical layer, players such as Zayo compete for fibre and wavelength demand. In the cloud‑connectivity layer, Megaport and Equinix Fabric offer software‑driven interconnection through partner and facility ecosystems. In the enterprise layer, regional carriers, SD‑WAN providers, and managed‑services firms assemble alternatives using the Internet underlay. These are overlapping but not identical categories.

Hyperscalers’ private backbones pose a different challenge. Cloud providers internalise inter‑region traffic and can blend transport and compute. Arelion is useful where it connects multiple clouds, enterprises, data centres, and networks that are outside a single corporation’s control. The more traffic stays inside a hyperscaler domain, the smaller some wholesale opportunities become. The more it is distributed across several, the more neutral wide‑area reach gains value.

IXs both complement and substitute for transit. Networks with enough traffic can peer directly with major destinations, reducing transit purchases. They still need reach to the rest of the Internet, transport to IX locations, and operational support. Arelion’s IX Connect and transit sit on both sides of this decision.

Arelion’s sustainable advantage is not that any single element is uncopiable. Fibre can be leased, routers bought, cloud on‑ramps joined, and DDoS systems deployed. It is the combined system of routes, peering relationships, customer prefixes, PoPs, optical paths, support, and reputation that takes years to build and continuous investment to keep relevant.

Connectivity Creates Resilience and Propagation Risk at the Same Time

A highly connected backbone has many path options and direct reach. It can avoid intermediate networks and shift flows during failure. The same position also increases the number of downstream organisations that can be affected by an error. Route leaks, filter mistakes, and capacity failures can propagate beyond direct customers because other networks depend on paths learned from AS1299.

Physical concentration has the same duality. Dense PoPs improve interconnection, yet a single facility failure can affect many services. A logically diverse topology can share the same fibre duct. DDoS infrastructure can protect customers, but a diversion‑policy mistake can move massive amounts of legitimate traffic. Scale amplifies both capability and blast radius.

Resilience depends on controls that are less visible than a map: maintenance discipline, staged configuration, route filters, RPKI validation, community management, optical protection, capacity headroom, scrubbing placement, out‑of‑band access, incident communication, and verified rollback. Arelion’s public tools and product pages show some of these, but the evidence pack does not contain a full failure history or independent audit.

Customers also carry responsibility. Multihoming, prefix design, routing policy, testing plans, and local access diversity determine how far the carrier’s resilience extends into the customer boundary. Buying from a Tier-1 does not remove the need to engineer for survival when that carrier is lost.

Sustainability Claims Require Denominator Discipline

Backbone growth consumes hardware, power, colocation, construction, and maintenance. Newer coherent optics can increase capacity per watt and per rack space. But if traffic and network scale outrun efficiency improvements, total energy use can still rise. Arelion publishes a sustainability framework, but the provided evidence lacks a full independently audited footprint for the entire network and supply chain.

Responsible treatment separates efficiency from absolute impact. A route refresh may lower energy per bit carried, while new routes may increase total equipment and power. Leasing infrastructure may shift some emissions and power use to a supplier but does not eliminate them. Without consistent scope, base year, and traffic data, broad claims of environmental improvement lack grounding.

Sustainability also matters strategically. Customers and pension‑capital owners may demand credible evidence on energy, procurement, and resilience. Operationally, higher optical‑equipment density, cooling, and power availability already shape where capacity can be placed. What to watch is whether reporting becomes detailed enough to tie network expansion to measured resource use.

What the Public Evidence Does Not Tell Us

The network‑scale numbers used in this briefing are primarily Arelion’s current self‑reporting. The research pack does not contain a route‑by‑route independent audit of the 80,000‑plus kilometres, a common definition for all PoPs, or a full inventory of owned versus leased infrastructure. There is no comparable traffic time‑series across competitors.

Connectivity claims are metric‑dependent. The “95% one‑hop” statement reflects Arelion‑reported reach in North America and Europe, not a guarantee for every destination, protocol, or point in time. Monthly performance reports are useful but operator‑selected. Procurement still requires route‑specific testing and contractual terms.

Product availability is also conditional. Ethernet supports up to 400 Gb/s, but not every location, access tail, and route can deliver 400G. AI Direct is a connectivity suite, not proof of AI customer revenue. The 6.1 Tb/s observed attack does not demonstrate contractual scrubbing capability. MEF/Mplify certifications support confidence in service definitions but do not certify every access tail or failure outcome.

The financial gap is larger. Current revenue, margin, debt, capex, valuation, and customer concentration are unpublished in the provided materials. One cannot build a complete economic model. The missing pieces should be noted explicitly, not filled by inference from route ranking, customer count, or the nature of the owner.

Even When Routing Identifiers Appear Stable, Refresh Continues

From the outside, AS1299 can look remarkably stable. The autonomous system number does not change, known peers remain visible, the Twelve99 technical name persists, and customers continue to receive global routing reachability. That stability is valuable, but it can also hide the amount of change required underneath. A backbone can swap out routers, optical platforms, line cards, software, power agreements, fibre paths, and facility attachments while keeping the same public routing identifier. The actual product is continuity achieved through controlled change.

Hardware generations set the first refresh clock. Router capacity must keep pace with increasing interface speeds, routing‑table size, telemetry, and security features. Chassis and line cards that were adequate during a 100G‑centric era become constraints when 400G and higher aggregation loads become normal. Whether a piece of hardware can still forward packets is not the only test; spares availability, manufacturer support, power density, software compatibility, and the cost of running multiple generations in parallel can make technically functional gear economically obsolete.

Optical systems create a second refresh clock. Coherent technology can dramatically increase the usable capacity of an existing fibre pair, but an upgrade is not a universal software switch. Distance, fibre characteristics, spectrum, amplifiers, ROADMs, open‑line design, and operational qualification determine what can actually be carried on a specific route. Whether Arelion can offer a higher‑speed service depends on per‑route engineering, not just on a module’s nominal speed.

Software creates a third refresh clock. Routing platforms receive security fixes, protocol changes, new features, and operational modifications. A stable network may avoid the latest version until it is well proven, but staying on old releases too long increases support and security risk. Refreshes must be staged so that the fix itself does not cause a larger availability problem. On a Tier-1 backbone, a single policy or control‑plane mistake can affect more than one customer, so change‑management quality matters as much as the code itself.

Routing standards and operational practice create a fourth clock. RPKI adoption, filter expectations, peering policy, BGP community conventions, and routing security evolve even when the ASN does not. A network that operated well ten years ago cannot assume that the same set of controls is sufficient today. Customers and peers expect stronger validation, more transparent routing policies, and better response to leaks and hijacks. Maintaining trust requires not just capacity but process refresh.

Facilities have a slower but important refresh cycle. Even if a PoP stays in the same building, the power density, cooling, cross‑connect economics, and surrounding peer mix change. A historically important location may no longer be the best place for new high‑capacity hardware, yet the cost of moving existing customer and peer attachments is significant. New data‑centre campuses can draw traffic away from legacy carrier hotels. The network must reach new demand centres without eroding the connection density of older locations.

Customer geography also changes. Enterprise applications shift to cloud regions; content providers deploy more edge; AI clusters locate in data‑centre markets with available power; wholesale customers build new in‑country networks. Arelion can keep the same PoP count and still lose relevance if it is further from key demand. Conversely, a small number of strategic route additions can greatly increase the value of the existing backbone.

Security capacity has a particularly uneven demand curve. Normal traffic can be sized to typical peaks, but DDoS infrastructure must handle rare events. The largest attack seen one year can become ordinary a few years later. Designing scrubbing only for average load would be limited public evidence; holding unlimited idle capacity would destroy economics. What is needed is a distributed headroom and diversion flexibility that can absorb a reasonable envelope of attack sizes while using the same infrastructure efficiently during normal conditions.

Access partners have their own lifecycles. Regional carriers improve in quality, consolidate, change ownership, retire products, or fall behind. The best enterprise access of three years ago may not be the best today. Global service quality depends on ongoing supplier evaluation and the ability to move new orders—and, where possible, existing tails—to better options. The partner count demonstrates breadth; the refresh effort is about deciding which relationships will carry forward the next cycle of traffic.

Contracts also age. A long‑term capacity right that looked attractive when signed can appear expensive compared with market prices a few years later. Short‑term contracts preserve flexibility but increase renegotiation and repricing exposure. The expiry dates of facilities, fibre, and supplier agreements may not align with customer contracts. The carrier must manage the physical and contractual clocks together so that a gap in one does not strand the economics of the other.

People create a less visible refresh requirement. A global backbone depends on knowledge about routing policy, optical characteristics, past failures, supplier behaviour, and customer exceptions. Some of that can move into inventory systems and automation; some remains embedded in the organisation. Retirement, reorganisation, and outsourcing can remove context that is needed only during rare faults. A stable network needs not just spare hardware but succession, documentation, and training.

Automation does not remove these refresh cycles. It can raise inventory accuracy, speed delivery, and make dependencies visible, but it also raises the stakes of getting the data right. A wrong location ID, an out‑of‑date path, or a misapplied supplier boundary can propagate faster through an automated workflow than through a manual one. The more Arelion leans into standardisation and automation, the more important it becomes to validate the underlying data model.

The commercial conclusion is that refresh investment must not be judged only by visible expansion. Replacing line cards before they fail, adding headroom on existing routes, moving customers off correlated paths, and upgrading the security substrate protect existing revenue without adding a new country or PoP. An owner who focuses only on conspicuous growth may under‑invest; an operator who treats every preventive refresh as automatically necessary may over‑invest. Refresh decisions need to be tied to specific failure modes, capacity constraints, customer contracts, and measurable operational improvement.

The question “Is AS1299 still Tier‑1?” is therefore too narrow. The routing relationships can be maintained while other parts of the service improve or deteriorate. A broader set of tests ask whether the network keeps pace with traffic growth, whether physical diversity is real, whether trust with peers and customers continues, whether security capacity keeps up with threat, and whether capital is being placed before technical debt becomes visible as an outage.

Arelion’s long history is both evidence and burden. More than 30 years of operation shows that the network has survived multiple hardware, traffic, and ownership cycles. It also means that inside the network are decisions made under different technology generations and demand environments. A refresh is about keeping what is valuable and replacing what no longer fits. The AS number can stay constant only because much underneath it does not.

Buyers Should Verify Service Boundaries, Not the Tier-1 Label

For a customer, useful due diligence begins after confirming that AS1299 is Tier‑1. The next questions are where Arelion’s direct control starts and stops for the service being bought. A transit port inside an Arelion PoP, a managed enterprise tail via a regional partner, a wavelength over an Arelion route, and Cloud Connect each have different failure domains. Operationally distinct contracts can live behind the same brand and backbone.

Physical boundaries should be explicit. One needs to know whether the handoff is inside an Arelion PoP, a third‑party data centre, a customer facility, or a partner location. If a local access tail is included, identify who owns it, how it enters the building, and whether a second line genuinely uses a different physical path. Two circuit IDs are not proof of diversity; what matters is the separation of ducts, building entries, facilities, power, and upstream path.

Routing boundaries need separate checking. A transit customer controls its ASN and can use communities, preference, and multihoming. A DIA customer delegates more edge policy. An Ethernet customer may receive a Layer‑2 service whose internal routing is invisible. A Cloud Connect customer reaches an on‑ramp where a different administrative domain begins. These differences determine who can change what during an incident and should be documented at procurement time.

Capacity should be verified at the boundary being purchased. A backbone that supports 400G does not mean a local port, IX, cloud on‑ramp, or access carrier can offer the same speed. Buyers for AI or data‑centre use must separate core network capability from the speed they can actually order at both ends. Route qualification, port inventory, and protection options matter more than a generic “400G‑ready” description.

Latency claims also need route‑specific definition. Arelion’s connectivity can reduce intermediate networks, but the path with fewest AS hops is not always the physically shortest or lowest‑latency. BGP policy, fibre distance, metro access, and destination placement all matter. Latency‑sensitive customers should measure from the relevant point and understand whether the contracted service uses traffic engineering, protection paths, or normal Internet routing.

Availability promises are similarly bounded. The fact that Arelion publishes different SLAs for backbone, inter‑PoP, and end‑to‑end services already shows why a single headline number is limited public evidence. Customers need to know what is in scope, how planned maintenance is treated, what exclusions apply, how it is measured, and what happens when it is missed. A high percentage can still permit a critical outage if the measurement window or remedy terms do not match application loss.

DDoS procurement requires looking separately at detection, diversion, scrubbing, and emergency blackholing. The ability to observe terabit‑scale attacks is evidence of network visibility, but the customer needs to know its own protection model, activation conditions, targeted prefixes, clean‑traffic return, and escalation. Out‑of‑scope attacks, such as application‑layer, that require different controls are equally important.

Cloud Connect needs a three‑party operating model. Arelion provides the carrier path; the cloud provider controls the interface and internal fabric; the customer controls accounts and routing. Sorting out identifiers and contact points before an incident makes fault isolation easier. The value of a private connection is that it can improve path and control; it does not turn three administrative domains into one.

Partner‑delivered access deserves extra scrutiny because quality is hard to standardise globally. Buyers should determine whether Arelion monitors the local tail, receives preventive alarms, can drive supplier escalation, and can obtain physical‑path information. What happens when a regional carrier repeatedly misses repair targets is as important as the initial delivery. The value of a global contract is high only when the prime contractor has enough data and commercial leverage to change the outcome, not just pass the ticket.

Operational transparency should be evaluated before an outage, not after. The looking glass, performance reports, and MyArelion are useful, but customers need to know which metrics are visible on their own services, how frequently they update, and whether evidence can be exported. When a performance dispute arises, common timestamps, path observations, and incident logs are more valuable than a generic status dashboard.

Change management is also part of the service boundary. Customers should understand what they can change themselves via portal or BGP community, what requires carrier approval, and what changes Arelion can initiate during maintenance or security response. Delegated control is useful only when the behaviour, blast radius, and rollback are understood—especially when multiple logical services share a single physical port.

Multihomed customers should not assume redundancy because two contracts are signed; they should test it. Route preference can keep traffic on a degraded or failed path longer than expected, customer prefixes can be filtered accidentally, and inbound and outbound routes can behave differently. Regular failover testing demonstrates whether AS1299 and an alternate carrier work as designed.

Optical diversity requires the same rigour. A protected wavelength may use two fibre paths, but the buyer still needs to know how those paths are defined and where they reunite. Two circuits with different terminations can provide a stronger boundary than two channels that share the same facility. If the facility itself is a critical failure domain, application design must also span multiple data centres.

At renewal time these assumptions should be re‑verified. A service bought three years ago may now be delivered over a different access supplier, a different facility, or a different network configuration. New cloud regions and data centres may also create better handoff points. The carrier may have added direct paths and changed the optimal design. Treating renewal as a price negotiation alone misses the chance to re‑baseline resilience design.

For BTW readers, this verification exercise shows the practical meaning of Tier‑1. AS1299 gives Arelion strong routing position, but customers buy services at specific boundaries. The backbone’s value materialises only when routing policy, physical infrastructure, partner access, security, and operations work as a system at that boundary. The Tier‑1 label reduces one kind of dependency; it does not remove the others.

The Central Judgement

Arelion illustrates what a Tier-1 carrier has become. The settlement-free reach of AS1299 remains a core networking position, but the commercial product is a multi-layer service substrate. Fibre and optics provide paths and capacity, PoPs provide market access, BGP, communities, and routing security provide policy. Ethernet, cloud, IX, and managed services create different boundaries. DDoS systems turn backbone visibility into a security capability. Operations and customer support turn all of these into something enterprises and carriers can buy.

The barrier to entry is historical accumulation. Arelion inherited decades of routes, facilities, peering relationships, and operational knowledge from the Telia Carrier lineage. Polhem Infra ownership provided a focused infrastructure investor and a new brand without replacing the technical identity. Twelve99 persists because network memory outlasts corporate marketing.

The same history also creates obligations. Tier-1 status does not freeze a network at the moment of attainment. Traffic prices change, attack peaks grow larger, clouds internalise transport, AI creates new corridor demand, optics get faster, and customers expect more automation. Arelion must refresh the substrate while preserving the routing trust that makes AS1299 valuable.

The evidence confirms that Arelion is active, globally significant, and technically broad. It does not confirm current margins, leverage, or how quickly AI-related announcements will turn into sustained revenue. The most useful conclusion is therefore specific: a Tier-1 today is a routing relationship wrapped in a capital-intensive system of fibre, optics, packet, security, and operations, whose quality must be proven repeatedly at the service boundary rather than granted once by a label.

Indicators That the Backbone Is Strengthening

Arelion’s next phase should be judged by the evidence of translation, not by slogans. The company already possesses a mature routing identity, self-reported global scale, and a broad product list. What remains open is whether new routes translate into real demand, whether value-added services compensate for falling transit prices, and whether resilience keeps pace with connectivity. The indicators below tie technical activity to commercial and operational outcomes.

Peer and route stability

Tier-1 status depends on maintaining settlement-free relationships. A significant externally visible change in adjacencies, path lengths, or route propagation from AS1299 would be an early sign that the economics of interconnection have shifted. Looking glasses and external BGP observations can show change but cannot explain the commercial conditions behind it.

Lit capacity, not announced capacity

New 400G products, optical refresh, and route announcements matter only when ports are lit, traffic grows, and customers contract. Track the geographic reach of qualified 400G, the completion of Nordic and North American routes, new PoPs, and named customer adoption. Sustained announcements without lit evidence weaken the AI and capacity‑scaling narrative.

Product mix beyond transit

Traffic volume can rise while per-bit price falls. Evidence that existing customers add SecureConnect, Cloud Connect, IX Connect, managed optical networking, and Ethernet would show Arelion monetising integration, not just commodity transit. Because the company does not publish revenue by product, customer and service announcements are the best currently available proxy.

Access‑partner performance

Growth through about 450 access partners extends reach but introduces variability in delivery and repair quality. Watch the gap between direct-PoP and end-to-end outcomes, local-tail delivery times, repeated regional faults, and whether Arelion discloses tighter partner qualification. The published SLA differential already shows that last-mile changes the risk structure.

DDoS scale and mitigation evidence

The 2026 report shows the attack landscape Arelion observed. Going forward, watch not just a single headline number but how peak volumes, botnet composition, mitigation time, customer impact, and scrubbing placement evolve. Observing a larger attack is not the same as evidence that protection has improved.

Capital and governance signals

Private ownership limits direct financial analysis. Track management changes, completed route projects, hiring, supplier agreements, disclosed funding events, and the investment posture of Polhem Infra. A slowdown in deployment or an increase in partner reliance could signal capital discipline, demand uncertainty, or both.

Five Evidence-Based Scenarios

Wide-area AI demand becomes a sustainable growth engine

Neo-clouds, enterprises, and research users distribute data and workloads across multiple facilities. Arelion lights 400G EVPL and wavelengths over named corridors, and AI Direct becomes a measurable source of traffic and customer expansion.

Traffic rises but the economics get tougher

AS1299 carries more bits, yet transit prices fall faster than utilisation rises. Arelion remains technically important while profitability becomes more dependent on cloud, security, Ethernet, and managed optics.

Security integration improves customer retention

Customers buy transit or DIA together with automated mitigation; SecureConnect plays a larger role in renewals and incident response. Success shows up in attachment rates, customer examples, and evidence that availability was preserved during large-scale attacks.

Hyperscalers internalise more transport

Cloud providers carry more traffic inside their own private backbones, reducing some wholesale demand. Arelion retains value for enterprises that need neutral multi-cloud and inter‑data‑centre paths, but corridor economics become more selective.

A correlated shock in routes, facilities, or capital exposes concentration

A peering change, fibre incident, facility failure, or refresh delay affects multiple services at once. The strategic outcome depends on physical diversity, capacity headroom, incident transparency, and the owner’s willingness to invest in recovery.

Practical Implications for Stakeholders

Network operators should test routing policy, community behaviour, and multihoming rather than treat Tier-1 status as redundancy. Enterprise buyers should map the direct and partner segments for each service. Cloud teams should define the boundary between Arelion and the on‑ramp provider; security teams must separate observed attack volumes from contracted mitigation capacity. Investors and owners should tie route expansion to utilisation and revenue. Regulators and resilience planners should examine facility, cable, and AS concentration, not treat ownership as the only form of control.

The Control, Incentives, and Decisions That Will Shape AS1299's Next Decade

Arelion’s management challenge is not a choice between “network” and “business.” The network itself is the business, and every technical decision locks in years of capital, supplier, and customer commitments. The company’s control map is distributed: Polhem Infra controls ownership and investment; management decides capital allocation and product priorities; engineering and operations control route and capacity changes; peers determine settlement-free relationships; facilities and access partners control portions of the physical path; customers control prefixes, demand, and multihoming; cloud providers control the far side of the on‑ramp.

The strategy succeeds only when these authorities are aligned.

Decision 1: Fund physical diversity before advertising reach

Arelion should measure route diversity not just by logical topology but across cable, duct, landing station, facility, and power layers. A new PoP or new path can increase apparent reach while sharing existing capacity and hidden failure domains. The hard-to-reverse risk is locking in infrastructure that cannot deliver the resilience sold at higher layers.

Management should embed evidence of physical diversity into capital approval and customer design. It will increase cost and may slow some deployments. At the same time, it reduces the probability that a single incident will silence multiple revenue products and protects the credibility of the global map. Even a carrier with a mixed ownership model can earn trust if it can demonstrate it controls the failure boundaries.

Decision 2: Monetise direct reach while protecting interconnection neutrality

AS1299’s barrier to entry depends on peers and customers continuing to see it as a trustworthy exchange partner. Product growth must not distort routing policy in ways that weaken that trust. Commercial pressure can encourage overly aggressive preference, concentration on a few large customers, or under-investment in paths that are topologically necessary but contribute less direct revenue.

Management should clearly separate settlement-free peering policy, customer transit policy, security interventions, and product traffic engineering. The second-order effect is easier incident diagnosis and greater counterparty trust. The third-order risk of failure is the loss of direct relationships that cannot be recovered just by buying more fibre.

Decision 3: Require lit evidence for the AI portfolio

AI Direct gives Arelion credible language to describe high-capacity inter‑data‑centre demand. It can also become a container that re‑labels standard circuits without new economics. Management should manage the portfolio with route-specific availability, lit ports, named use cases, utilisation, protection requirements, and renewal behaviour.

This discipline prevents capital chasing market enthusiasm without customer evidence. It also refines product design because training data, inference traffic, replication, and cloud connectivity have different latency and security needs. The hard-to-reverse risk is overbuilding corridors or optical capacity that does not attract the demand that was projected.

Decision 4: Turn security evidence into an operational feedback loop

The DDoS report gives Arelion a valuable view of attack traffic. Management should connect that publication to capacity planning, customer design, scrubbing placement, false-positive verification, and post-incident learning. Reporting only peak volumes evaluates spectacle; reporting mechanisms and outcomes improves the service.

A stronger loop distinguishes observed attacks, mitigated attacks, customer impact, and the role of blackholing. Customers can calibrate risk; owners can assess whether security investment is protecting revenue. Without that evidence, SecureConnect risks commoditising into a label over the long term rather than a demonstrated operational capability.

Decision 5: Make partner reach governable

Access partners are essential to global enterprise delivery, but accountability can split across contracts. Arelion should treat partner selection, service catalogue, incident data, diversity, and exit rights as part of product design. Customers need to know which segments are direct, which are managed, and which are simply coordinated.

The second-order effect of strong governance is that it extends the addressable market without building the same footprint itself. The third-order effect of weak governance is that a single regional supplier can damage the reputation of the global backbone. Some failures will remain outside Arelion’s physical control, but they need not remain outside its evidence and escalation control.

Decision 6: Disclose enough economics to sustain institutional trust

Private ownership gives Arelion flexibility, but customers, partners, and infrastructure counterparties cannot assess leverage, capex, or customer concentration. Full public-company disclosure is not needed to improve trust. Management and the owner can disclose consistent indicators on operational investment, capacity utilisation, resilience, and sustainability without revealing contract-confidential data.

Better disclosure raises the meaning of route announcements and reduces the temptation to use connectivity as a proxy for financial health. It can tie strategy to comparable outcomes and also discipline internal capital allocation. The risk of staying opaque is that, during market stress, counterparties will make their own, often harsher, assumptions.

Second-Order Effects of Successful Execution

If Arelion turns long-term capital into physically diverse capacity, protects peer trust, and adds value-added services, AS1299 becomes more than a commodity transit path. It can become a neutral wide-area substrate for clouds, enterprises, carriers, and distributed AI infrastructure. That position can improve customer retention, refine capacity forecasting, and make security investment easier to justify economically.

Success also affects the surrounding ecosystem. Access partners gain traffic and standardisation pressure; data centres gain interconnection value; cloud providers gain enterprise reach; IXs gain remote entities; customers gain an alternative to concentrating transport with a single hyperscaler. Because the benefits are distributed, no single entity controls the full outcome.

Third-Order Effects of Failure

A significant routing-policy mistake or hidden physical concentration can create damage beyond a single product. Customers shift traffic, peers reassess trust, security services lose credibility, and the owner faces larger capital demands at the same time revenue is at risk. Highly connected infrastructure can propagate reputational loss as fast as it can propagate packets.

Long-term capital constraint has a slower but persistent effect. Delayed optical and PoP refreshes lengthen paths, increase partner dependence, and make the network less attractive for high-capacity customers. If direct relationships and anchor customers migrate, rebuilding can take years.

Hard-to-Reverse Risks

The hardest to recover from are loss of settlement-free peering relationships, long-term lock-in to physically correlated paths, under-investment across a whole hardware generation, dependency on a single large customer segment, and a security incident that exposes weak operational control. Brand damage can be repaired; lost route position or uneconomic long-term infrastructure contracts can remain.

Management should therefore distinguish between product experiments that are reversible and network commitments that are hard to unwind. A portal feature can be changed; a fibre right, landing-station dependency, or optical architecture constrains the company for years. A pricing campaign can end; a peering relationship damaged by policy abuse does not return on demand.

Final Management Judgement

Arelion’s strategic asset is not the Arelion brand alone or the Twelve99 hostname alone. It is the coordinated system that makes AS1299 work: the interconnection relationships, the physical paths, the optical capacity, the packet policies, the security, the operations, and the trust of customers and peers. Polhem Infra and management do not start from a blank sheet; they inherit that system. The task is to refresh it without breaking the trust that accumulated under previous identities.

The decisive test is disciplined translation: new routes must become diverse and available paths; new ports must become customer traffic; AI positioning must become lit services; DDoS visibility must become sustained availability; partner reach must become governable delivery; private capital must become timely refresh that does not hide risk. When those translations hold, Tier-1 status retains its economic meaning. When they do not, the label will outlast the advantage it once described.