The Portugal-Spain match at the 2026 World Cup left more than sporting tension behind. It also set a new all-time traffic record in DE-CIX’s global interconnection ecosystem: 28.4 Tbit/s, registered on 06/07/2026 at 19:35 UTC, 21:35 Spanish peninsular time. The figure confirms a trend that has been building for years: major live events are no longer just a television matter, but a stress test for networks, data centers, cloud platforms, CDNs and interconnection operators.
The data has a clear technical reading. Internet traffic does not grow only through the gradual accumulation of users, devices and applications. Increasingly, it also grows through highly concentrated peaks in time. Millions of people connect almost simultaneously, from smart TVs, smartphones, tablets and computers, to consume the same live content. That synchronization changes the way infrastructure is designed: it is no longer enough to support a reasonable average traffic load; platforms must be ready to absorb massive bursts without degrading the user experience.
Live streaming changes traffic patterns
For years, much of the growth in Internet traffic has been linked to predictable trends: more remote working, more video calls, more e-commerce, more on-demand audiovisual content and more cloud services. Stackscale had already analyzed this growth in previous articles on the rise in global Internet traffic between 2020 and 2021 and on the increase in global traffic in 2024. But sports streaming introduces a different variable: simultaneity.
A series watched on demand, a software update or an enterprise backup spreads the load more evenly over time. A World Cup knockout match does not. The peak appears within minutes, often just before kick-off, during key plays or at half-time, when searches for statistics, social media activity, highlights, chats and second-screen applications also surge.
DE-CIX also points to a factor that will gain weight in the coming years: content generated or enriched by Artificial Intelligence. Automated highlights, personalized recommendations, real-time statistics and interactive experiences add new layers of traffic on top of the main broadcast. Video remains the major consumer of bandwidth, but a whole set of services is growing around it, also requiring low latency, availability and responsiveness.
For companies, the lesson goes beyond sport. Any digital business with campaigns, launches, sales periods, online events, live broadcasts, auctions, ticket sales or data-intensive services may face similar patterns, even if at a different scale. Infrastructure must be designed with the real peak hour in mind, not just the monthly average.
Internet exchanges, CDNs and latency: the invisible side of streaming
When a user clicks “watch live”, they expect the stream to load quickly, remain stable and have low delay. But behind that action lies a complex chain: access providers, backbone networks, cloud platforms, data centers, CDNs, streaming operators and Internet Exchange Points.
Internet Exchange Points, or IXPs, allow different networks to exchange traffic directly, without forcing data to travel through unnecessary routes. Stackscale had already explained in its article on Internet Exchange Points how this direct interconnection helps improve latency and bandwidth. It also published specific pieces on DE-CIX, one of the world’s largest Internet Exchange Point operators, and on AMS-IX, the Amsterdam Internet Exchange.
The logic is simple: the closer content is to the user and the fewer hops it has to cross, the better the experience. That is why CDNs distribute copies of content across multiple locations, and why well-connected data centers are so important. In a global event such as the World Cup, it is not only about having powerful servers, but about placing capacity in the right locations and connecting it to the right networks.
Latency, also analyzed on the Stackscale blog, is one of the factors that most affects perceived quality. In traditional video streaming it may seem less critical than in gaming, trading or industrial applications, but in live broadcasting it makes a difference: delay compared with the original feed, buffering, quality changes, start-up times and synchronization with statistics or interactive experiences.
The physical infrastructure of the Internet also comes into play. Submarine cables, metropolitan networks, fiber optics, meet-me rooms in data centers and links with operators are all part of an architecture the user does not see, but which sustains the digital experience. Massive streaming does not live in “the cloud” as an abstract concept: it lives in racks, routers, switches, fiber links, interconnection ports, storage systems and distributed computing platforms.
What companies should learn from these traffic peaks
The 28.4 Tbit/s record does not mean that every company has to size its platform like a global streaming operator. But it does leave a useful idea: peaks do not forgive poorly designed architectures.
In e-commerce, digital media, SaaS, education platforms, gaming, fintech or B2B services with high-demand windows, infrastructure must combine several layers. The first is computing capacity: enough CPU, memory and storage to handle the load. The second is networking: bandwidth, routing, redundant connectivity and low latency. The third is application architecture: caching, load balancing, queues, properly sized databases, separation of critical services and observability.
Stackscale recently published a guide on how to prepare infrastructure for traffic peaks without the website going down, with a very practical idea: before adding resources without a clear plan, it is worth reviewing the frontend, database, media, caches, physical limits and bottlenecks. More servers do not always solve a poor design. Sometimes the problem lies in an inefficient query, storage saturated by IOPS, a poorly configured CDN or a database concentrating too much load.
This is where private cloud and bare-metal play a relevant role. For critical workloads, latency-sensitive services or applications with intensive and predictable consumption, dedicated resources help reduce the risk of noisy neighbors and provide greater control over CPU, memory, network and storage. It is not always the right option for every workload, but it fits scenarios where sustained performance, cost predictability and isolation matter.
In European environments, data sovereignty is also part of the equation. Hosting infrastructure in data centers within Europe, with the right connectivity and close operational support, makes technical, contractual and regulatory control easier. This is especially important in sectors that work with sensitive data, critical services, intellectual property or compliance requirements.
High availability must also be approached realistically. It is not enough to replicate virtual machines. RPO, RTO, service dependencies, backup strategy, disaster recovery, monitoring and regular testing must all be defined. A traffic peak can expose failures that remain hidden for weeks: connection limits, firewall saturation, insufficient load balancer capacity, slow storage or network paths that are not ready to absorb sustained traffic.
The World Cup shows the Internet at its most demanding: millions of users, simultaneous consumption, high-quality video and added services around the broadcast. For companies, the lesson is closer than it might seem. Infrastructure should not be designed only to work when everything is calm, but to keep responding when many users do the same thing at the same time.
Frequently asked questions
What does a 28.4 Tbit/s Internet traffic peak mean?
It means that, at a specific moment, DE-CIX’s global interconnection ecosystem moved 28.4 terabits per second. It is a measure of aggregated traffic and reflects the pressure that massive live events can place on Internet infrastructure.
Why does sports streaming generate such large traffic peaks?
Because it concentrates millions of users connecting at the same time. Unlike on-demand video, a live match does not spread consumption over hours or days, but concentrates demand in very specific time windows.
What role do Internet Exchange Points play?
Internet Exchange Points allow the direct exchange of traffic between networks, operators, CDNs, cloud providers and digital platforms. They help reduce unnecessary routes, improve latency and increase network efficiency.
How can a company prepare for traffic peaks?
It should review the full architecture: application, database, cache, load balancing, network, storage, monitoring, backups and scaling capacity. For critical workloads, a private cloud or bare-metal environment can provide dedicated performance, control and predictability.
