Christopher Anvil

A Tier 1 network is an Internet Protocol (IP) network that can reach every other network on the Internet solely via settlement-free interconnection (also known as settlement-free peering).^[1][2] Tier 1 networks can exchange traffic with other Tier 1 networks without paying any fees for the exchange of traffic in either direction.^[3] In contrast, some Tier 2 networks and all Tier 3 networks must pay to transmit traffic on other networks.^[3]

There is no authority that defines tiers of networks participating in the Internet.^[1] The most common and well-accepted definition of a Tier 1 network is a network that can reach every other network on the Internet without purchasing IP transit or paying for peering.^[2] By this definition, a Tier 1 network must be a transit-free network (purchases no transit) that peers for no charge with every other Tier 1 network^[4][5] and can reach all major networks on the Internet. Not all transit-free networks are Tier 1 networks, as it is possible to become transit-free by paying for peering, and it is also possible to be transit-free without being able to reach all major networks on the Internet.

The most widely quoted source for identifying Tier 1 networks is published by Renesys Corporation, but the base information to prove the claim is publicly accessible from many locations, such as the RIPE RIS database,^[6] the Oregon Route Views servers, Packet Clearing House, and others.

It can be difficult to determine whether a network is paying for peering or transit, as these business agreements are rarely public information, or are covered under a non-disclosure agreement. The Internet peering community is roughly the set of peering coordinators present at the Internet exchange points on more than one continent. The subset representing Tier 1 networks is collectively understood in a loose sense, but not published as such.

Common definitions of Tier 2 and Tier 3 networks:

• Tier 2 network: A network that peers for no charge with some networks, but still purchases IP transit or pays for peering to reach at least some portion of the Internet.
• Tier 3 network: A network that solely purchases transit/peering from other networks to participate in the Internet.

Since approximately 2010, this hierarchical organization of Internet relationships has evolved. Large content providers with private networks and CDNs, like Google, Netflix, and Meta, have greatly reduced the role of Tier 1 ISPs and flattened the internet topology since the content providers interconnect directly with most other ISPs, bypassing Tier 1 transit providers.

The original Internet backbone was the ARPANET when it provided the routing between most participating networks. The development of the British JANET (1984) and U.S. NSFNET (1985) infrastructure programs to serve their nations' higher education communities, regardless of discipline,^[7] resulted in the NSFNet backbone by 1989. The Internet could be defined as the collection of all networks connected and able to interchange Internet Protocol datagrams with this backbone. Such was the weight of the NSFNET program and its funding ($200 million from 1986 to 1995)—and the quality of the protocols themselves—that by 1990, when the ARPANET itself was finally decommissioned, TCP/IP had supplanted or marginalized most other wide-area computer network protocols worldwide.

When the Internet was opened to the commercial markets, multiple for-profit Internet backbone and access providers emerged. The network routing architecture then became decentralized and this meant a need for exterior routing protocols: in particular, the Border Gateway Protocol emerged. New Tier 1 ISPs and their peering agreements supplanted the government-sponsored NSFNet, that program being officially terminated on April 30, 1995.^[7] The NSFnet-supplied regional networks then sought to buy national-scale Internet connectivity from these now-numerous private long-haul networks.

A bilateral private peering agreement typically involves a direct physical link between two partners. Traffic from one network to the other is then primarily routed through that direct link.

A Tier 1 network may have various such links to other Tier 1 networks.^[8][9][10] Peering is founded on the principle of equality of traffic between the partners and as such disagreements may arise between partners in which usually one of the partners unilaterally disconnects the link in order to force the other into a payment scheme. Such disruptive de-peering has happened several times during the first decade of the 21st century. When this involves large-scale networks involving many millions of customers this may effectively partition a part of the Internet involving those carriers, especially if they decide to disallow routing through alternate routes. This is not largely a technical issue but a commercial matter in which a financial dispute is fought out using the other party's customers as hostages to obtain a better negotiating position. In the worst case, single-homed customers of each network will not be able to reach the other network at all. The de-peering party then hopes that the other network's customers will be hurt more by the decision than its own customers which may eventually conclude the negotiations in its favor.^[11][12] Lower tier ISPs and other parties not involved in the dispute may be unaffected by such a partition as there exist typically multiple routes onto the same network. The disputes referenced have also typically involved transit-free peering in which one player only exchanged data with the other that involved each other's networks—there was no data transiting through the other's network destined for other parts of the Internet. By the strict definition of peering and the strict definition of a Tier 1 network, a Tier 1 network only peers with other Tier 1 networks and has no transit routes going anywhere. More practically speaking, Tier 1 networks serve as transit networks for lower tier networks and only peer with other Tier 1 networks that offer the same services on an adequate scale—effectively being "peers" in the truest sense of the word.^[13]

More appropriately then, peering means the exchange of an equitable and fair amount of data-miles between two networks, agreements of which do not preclude any pay-for-transit contracts to exist between the very same parties. On the subject of routing, settlement-free peering involves conditions disallowing the abuse of the other's network by sending it traffic not destined for that network (i.e. intended for transit). Transit agreements however would typically cater for just such outbound packets. Tier 1 providers are more central to the Internet backbone and would only purchase transit from other Tier 1 providers, while selling transit to providers of all tiers. Given their huge networks, Tier 1 providers often do not participate in public Internet Exchanges^[14] but rather sell transit services to such participants and engage in private peering.^[15] Colocation centers often host private peering connections between their customers, internet transit (tier 1) providers and cloud providers.^[16][17]

In the most logical definition, a Tier 1 provider will never pay for transit because the set of all Tier 1 providers sells transit to all of the lower tier providers everywhere, and because

As such, by the peering agreement, all the customers of any Tier 1 provider already have access to all the customers of all the other Tier 1 providers without the Tier 1 provider itself having to pay transit costs to the other networks. Effectively, the actual transit costs incurred by provider A on behalf of provider B are logically identical to the transit costs incurred by provider B on behalf of provider A—hence there not being any payment required.

The traditional tiered hierarchy of the internet was challenged by the emergence of hyperscaler content and cloud providers. These providers have become major sources of global internet traffic, and in response, have built their own global network infrastructure that operates in parallel to, and in many cases bypasses, the traditional Tier 1 backbones.^[18]

A few hyperscale companies—specifically Google (Youtube, Cloud), Meta (Facebook, Instagram), (AWS, Prime Video), Microsoft (Azure, Xbox), and Netflix—are the dominant source of this traffic, accounting for nearly 75% of all used international bandwidth as of 2024.^[19][20] This marks a significant shift from the earlier internet, where traffic sources were more diffuse. The concentration of traffic provided an economic incentive to build their own infrastructure. When the majority of an end-user's traffic is destined for only a handful of networks, the value of a single, paid transit connection to the "entire internet" diminishes for an access ISP. It becomes more efficient for that ISP to peer directly with the major traffic sources, a dynamic that reduces the value of the core transit product offered by Tier 1s and fuels the rise of direct interconnection and on-net content caching.^[21]

In response to their traffic generation, hyperscalers have made substantial investments in building their own private global backbones, which in scale and sophistication can rival or exceed those of traditional Tier 1 ISPs.^[22] This infrastructure has several key components:

• Global Data Centers and Cloud Regions: AWS, Microsoft Azure, and Google Cloud Platform operate hundreds of data centers organized into dozens of geographically distinct "cloud regions" and "availability zones" around the world.^[23][24][25] These facilities form the core hubs of their networks.
• Private Fiber Backbones: These global data centers are interconnected by extensive, private fiber-optic networks. Traffic between regions on these backbones—for example, between two VPCs in an inter-region VPC peering connection—is encrypted and remains on the provider's private network, without traversing the public internet.^[26][27]
• Submarine Cable Investment: Historically, network builders would lease capacity on submarine communications cables owned by telecom consortia or Tier 1 providers. Today, content providers are the primary investors in new submarine cable construction, funding over $11 billion in new cables expected to enter service between 2023 and 2025.^[28][29] Specifically, Google is believed to operate the world's largest global backbone today, with at least 31 submarine cables^[30], most of which are privately owned. This shift from being the largest customers of capacity to the largest owners gives them control over their own supply and costs.^[31] This change in ownership economics has altered the traditional customer-supplier relationship, as hyperscalers now possess surplus fiber pairs on their cables that they can, in turn, sell or trade with other networks, including the Tier 1s from whom they once leased.

To deliver content to end-users with lower latency and cost, hyperscalers employ two primary strategies to bypass the traditional transit model and move their services closer to the network edge.

1. Extensive Peering: As outlined previously, hyperscalers pursue an open peering policy, establishing connections at IXPs and through private network interconnects (PNIs).^[32][33] This strategy is a response to the cost of IP transit bandwidth, which can differ significantly between regions.^[34][35] By peering directly with access ISPs, a hyperscaler offloads traffic that would otherwise have gone to a transit provider, improving performance and reducing operational expenditures.^[36]
2. On-Net Caching: Another strategy involves the content provider placing physical servers—caches—directly inside an access network's data centers. This localizes traffic, ensuring that popular content can be served to users without it needing to cross an external network boundary. The two most prominent examples are:
   • Netflix Open Connect (OCA): Netflix provides consumer ISPs that meet certain traffic thresholds with Open Connect Appliances (OCAs) at no charge.^[37] These are high-capacity storage servers that Netflix ships to the ISP. The ISP provides rack space, power, and a BGP session to the OCAs.^[38] During off-peak hours, Netflix populates the caches with its most popular regional content. When a subscriber in that ISP's network plays a video, the stream is served directly from the local OCA inside their own provider's network, significantly reducing the ISP's need to pull traffic from upstream transit links and improving the viewing experience.^[39]
   • Google Global Cache (GGC): Google operates a similar program for its high-bandwidth content, primarily YouTube and GCP. Google provides free GGC hardware to qualifying ISPs, who install the nodes in their network.^[40] The ISP establishes a BGP session with the GGC node and announces the IP address prefixes of its users.^[41] Google's traffic management systems then use this information, along with user DNS queries, to redirect users to the local GGC node for popular content. This keeps traffic local, reduces the ISP's operational costs, and lowers latency for users.^[42]

The structural changes to internet traffic flows have prompted a re-evaluation of the role and definition of Tier 1 networks. While the classic model is not obsolete, its dominance has been reduced, and its business focus has shifted to enterprises in response to the rise of hyperscale backbones.

A common discussion in network engineering circles is whether hyperscale networks like Google, Amazon, and Microsoft should now be considered Tier 1. In terms of physical infrastructure—global fiber backbones, submarine cable ownership, and points of presence—their networks are on par with, or in the case of Google greatly exceed, those of traditional Tier 1s.^[43][44][45] They are also largely "transit-free," meaning they do not purchase significant amounts of transit to reach destinations.

However, they do not fit the functional definition of a Tier 1 network because their primary purpose is not to sell universal IP transit to third parties.^[46][47] Their networks are built to serve their own applications, and their extensive peering is a distribution strategy designed to optimize that delivery.

Thus, instead of replacing Tier 1s, the hyperscalers have created a distinct, parallel category of internet backbone. The modern internet core is less of a strict hierarchy with Tier 1s at the apex, and more of a complex ecosystem where two types of large-scale global backbones (Tier 1 and Hyperscale) interconnect with each other and with thousands of smaller networks below them.

These networks are universally recognized as Tier 1 networks, because they can reach the entire internet (IPv4 and IPv6) via settlement-free peering. The CAIDA AS rank is a rank of importance on the internet.^[48]

Name	Headquarters	AS number	CAIDA AS rank[48]	Fiber route (km)	Peering policy
AT&T[49]	United States	7018	29	660,000[50]	AT&T Peering policy
Deutsche Telekom Global Carrier[51]	Germany	3320	17	250,000[52]	DTAG Peering Details
GTT Communications	United States	3257	4	232,934[53][54]	GTT Peering Policy
Liberty Global[55][56]	Netherlands[57]	6830	26	800,000[58]	Peering Principles
Lumen Technologies (formerly CenturyLink, formerly Level 3)[59][60][61]	United States	3356	1	885,139[62][63]	Lumen Peering Policy
NTT Communications (formerly Verio)[64]	Japan	2914	5	?	Global Peering Policy
Orange[65]	France	5511	11	495,000 [66]	OTI peering policy
PCCW Global	Hong Kong	3491	10	?	Peering policy
Tata Communications (prev. VSNL prev. Teleglobe)[67]	India	6453	6	700,000[68]	Peering Policy
Telecom Italia Sparkle (Seabone)[69]	Italy	6762	8	560,000	Peering Policy
Arelion (formerly Telia Carrier)[70]	Sweden	1299	2	70,000[71]	Arelion's IP Network Peering Policy
Telxius (Subsidiary of Telefónica)[72]	Spain	12956	15	65,000[73]	Peering Policy
Verizon Enterprise Solutions (formerly UUNET)[78]	United States	701	25	805,000[79]	Verizon UUNET Peering policy 701, 702, 703
Zayo Group (formerly AboveNet)[80]	United States	6461	9	196,339[81]	Zayo Peering Policy

While most of these Tier 1 providers offer global coverage (based on the published network map on their respective public websites), there are some which are restricted geographically. However these do offer global coverage for mobiles and IP-VPN type services which are unrelated to being a Tier 1 provider.

A 2008 report shows Internet traffic relying less on U.S. networks than previously.^[82]

A common point of contention regarding Tier 1 networks is the concept of a regional Tier 1 network. A regional Tier 1 network is a network which is not transit-free globally, but which maintains many of the classic behaviors and motivations of a Tier 1 network within a specific region.

A typical scenario for this characteristic involves a network that was the incumbent telecommunications company in a specific country or region, usually tied to some level of government-supported monopoly. Within their specific countries or regions of origin, these networks maintain peering policies which mimic those of Tier 1 networks (such as lack of openness to new peering relationships and having existing peering with every other major network in that region). However, this network may then extend to another country, region, or continent outside of its core region of operations, where it may purchase transit or peer openly like a Tier 2 network.

A commonly cited example of these behaviors involves the incumbent carriers within Australia, who will not peer with new networks in Australia under any circumstances, but who will extend their networks to the United States and peer openly with many networks.^{[citation needed]} Less extreme examples of much less restrictive peering requirements being set for regions in which a network peers, but does not sell services or have a significant market share, are relatively common among many networks, not just regional Tier 1 networks.

While the classification regional Tier 1 holds some merit for understanding the peering motivations of such a network within different regions, these networks do not meet the requirements of a true global Tier 1 because they are not transit-free globally.^[83]

This is a list of networks that are often considered and close to the status of Tier 1, because they can reach the majority (50+%) of the internet via settlement-free peering with their global rings. However, routes to one or more Tier 1 are missing or paid. Therefore, they are technically Tier 2, though practically something in between.

Name	Headquarters	AS Number	CAIDA AS Rank[48]	Reason
China Telecom	China	4134/4809	87	Purchases transit from Lumen/AS3356, Cogent/AS174, Verizon/AS701.
China Unicom	China	4837	181	Purchases transit from Cogent/AS174, Verizon/AS701, Lumen/AS3356, Orange/AS5511.
Singtel[84]	Singapore	7473	16	Purchases transit from Arelion/AS1299, Zayo/AS6461, Tata Communications/AS6453.
Cogent Communications (formerly PSINet)[85]	United States	174	3	No IPv6 routes to Hurricane Electric/AS6939
Hurricane Electric[86]	United States	6939	5	IPv4: Purchases transit from Arelion/AS1299 to reach GTT/AS3257, NTT/AS2914 and Cogent/AS174. IPv6: Lack of peering with Cogent/AS174.[87][88]
RETN[89]	United Kingdom	9002	12	Purchases transit from Lumen/AS3356 to reach Cogent/AS174 and NTT/AS2914.
Vodafone Carrier Services (formerly Cable & Wireless)[90]	United Kingdom	1273	13	Purchases transit from Arelion/AS1299 to reach AT&T/AS7018.[91]
Verizon Enterprise Solutions (formerly XO Communications)[92][93]	United States	2828	220	IPv6: Purchases transit from Cogent Communications/AS1239 to reach Vodafone (CW)/AS1273 and Telecom Italia Sparkle (Seabone)/AS6763.
Telstra[94]	Australia	4637	14	Purchases transit from Lumen/AS3356, Arelion/AS1299, Zayo/AS6461.
Comcast[95]	United States	7922	29	Purchases transit from Tata/AS6453

• Optical Carrier transmission rates
• Interconnect agreement
• Internet exchange point
• List of Internet exchange points