At first glance, asking whether Internet Exchange Points (IXPs) make the Internet faster may seem almost provocative at a peering forum. Surely the answer is simply ‘yes’?
However, an interesting panel discussion during the recent BKNIX Peering Forum (BPF) 2026 and ThaiNOG 8 showed that the answer depends on what we mean by ‘faster’, the type of IXP, where content is located, and how the Internet is evolving.
The panel, moderated by NSRC’s Philip Smith, brought together network operators, researchers, and academics. Participants included Romain Fontugne (IIJ Research Laboratory), Marc Bruyère (Airbus/SPIE ICS), Aftab Siddiqui (APNIC Foundation), Viraphan Samadi (Asian Institute of Technology), and Than Nye Htwe (NexGeNet).
Rather than resolving the question into a simple answer, the discussion highlighted how much the role of IXPs has changed since the early days of the Internet.
Define ‘faster’
A key theme was that ‘faster’ means different things to different users. For an online gamer, it means low latency. For a Netflix viewer, it means smooth, uninterrupted streaming. For researchers transferring large datasets, throughput may matter more than milliseconds. For a traveller using in-flight Wi-Fi, simply achieving a usable connection can feel fast.
As Romain noted, network performance is not a single metric. Latency, packet loss, jitter, congestion, reliability, and consistency all shape user experience.
Aftab offered a simple example. His son does not care about bandwidth or routing policies, he cares about whether his game has high latency.
In the end, users judge performance by experience, not infrastructure.
Throughout the session, discussion repeatedly returned to where the content is located.
In Australia, Aftab argued, IXPs clearly improve performance because much content is hosted locally. Traffic can stay within the same economy or region, reducing latency and transit costs.
However, his experience helping to establish Pakistan’s first IXPs showed a different reality. When most content is hosted overseas, local peering may not significantly improve user experience. Traffic still needs to cross international networks.
This reflects a broader shift. IXPs were once described as ways to ‘keep local traffic local’. Today, defining ‘local’ content is increasingly difficult. For example, a website serving Thai users may be hosted in a cloud region on another continent. At the same time, global platforms may cache content deep inside local networks. Locality has become blurred.
Not all IXPs are the same
Romain’s IIJ measurement research has highlighted another challenge: The term IXP now covers very different types of organizations. Some IXPs remain community-driven, non-profit facilities serving local networks. Others operate as large international platforms with members from multiple regions. Some are government-supported, while others are commercial.
Measured datasets show a clear split. Some IXPs remain highly regional, with most networks based in the same economy. Others have become international hubs connecting networks globally. Examples such as DE-CIX Frankfurt and Equinix Singapore arguably serve very different functions from traditional local IXPs, even though they share the same label.
The panel noted that the industry may eventually need new terminology because ‘IXP’ increasingly obscures these differences.
Beyond technical performance, IXPs can also speed up operational and security deployment.
Romain highlighted RPKI Route Origin Validation (ROV) as an example. Many IXPs implemented ROV on route servers before large transit providers deployed validation across their networks. By offering centralized validation, IXPs enabled security improvements to spread more quickly across the ecosystem.
In this sense, IXPs may make the Internet ‘faster’ not only by reducing latency, but by accelerating collaboration, innovation, and the adoption of best practices. This is a broader definition of speed.
While the technical argument is complex, the economic case is clearer. Several panellists stressed that peering remains significantly cheaper than buying transit. Than Nye Htwe described how joining BKNIX reshaped traffic flows for his network in Myanmar. By expanding peering with content providers, his organization improved user experience and reduced complaints about gaming and content access.
Viraphan shared a similar view from the academic sector: Cost remains a primary reason to connect to IXPs. The panel broadly agreed that when content is available at an exchange, peering offers clear economic advantages.
One topic that drew particular interest was Google’s shift towards ‘verified peering provider’ models. Although the panel did not reach consensus on Google’s motivations, it viewed the shift as reflecting Google’s scale rather than signalling the end of IXPs.
As Romain noted, hyperscale providers operate under very different constraints from most networks. Their size allows them to pursue connectivity strategies that smaller operators cannot. For most networks, IXPs remain one of the simplest and most cost-effective ways to reach users.
What is an IXP now?
A key insight came when Aftab reflected on how difficult it has become to define an IXP.
The original model was simple: Local networks exchanged local traffic. Today’s Internet is shaped by cloud platforms, distributed content delivery networks, regional peering fabrics, remote peering, and hyperscale infrastructure. Traffic flows no longer align neatly with geography.
IXPs increasingly function not just as local exchange points, but as platforms that support connectivity, resilience, security, and regional ecosystem development, so in this context, asking simply whether IXPs make the Internet faster may miss the bigger point.
Despite differing perspectives, the panel reached one clear conclusion: IXPs remain foundational infrastructure. Whether measured through latency, cost, resilience, security deployment, or connectivity, IXPs continue to play a critical role in the Internet ecosystem.
As Aftab concluded, while their role is evolving, IXPs remain one of the fundamental building blocks of Internet growth. The more useful question is no longer whether IXPs make the Internet faster, but how they will continue to adapt as the Internet evolves.
Watch the panel discussion in full:
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.