Alberto Caraballo
Heading into the final of the 2026 FIFA World Cup on July 19, Bank of America projected that a single match could account for as much as 7% of global internet traffic. It is a striking figure, but the more interesting question is rarely asked: where does all of that traffic physically change hands?
A large part of the answer is at Internet Exchange Points. And the last two World Cups have demonstrated it on three continents: live streaming has quietly rewritten what the internet demands from them.
Live is a different problem
For most of the streaming era, the internet’s defining challenge was distribution, and distribution is a problem you can solve in advance. A series episode, a software update, or a game download can be copied to caches close to the user days before anyone presses play. Demand spreads across evenings, weekends, and time zones. Content Delivery Networks were built for exactly this: move the content early, serve it locally, keep the load smooth.
Live inverts every one of those assumptions. Millions of people request the same content at the same instant, and that content cannot be pre-positioned because it does not exist until the moment it happens. The demand curve is not smooth; it is nearly vertical. Mauro Magrassi, CTO of the Milan Internet eXchange (MIX), has described the pattern with unusual precision: within minutes of kickoff, MIX registers a traffic increase of roughly 40%, around half a terabit of demand appearing almost at once.
The design consequence is fundamental. Infrastructure built for static content is dimensioned for averages. Infrastructure that carries live sport must be dimensioned for the single busiest minute of the year: the seconds before kickoff, the goal, the penalty shootout. That is precisely where an Internet Exchange earns its role. By letting content platforms, CDNs, and access networks hand off traffic directly, an IX absorbs the simultaneity locally instead of letting it detour through distant transit routes.
Three continents, one pattern
The evidence for this shift is not anecdotal. It is showing up, with remarkable consistency, in the traffic statistics of exchanges around the world.
In Brazil, IX.br, the largest public exchange ecosystem on the planet, reached an aggregate record of 50 Tbps earlier this year, with Sao Paulo alone concentrating 32 Tbps. More telling than the record is the planning behind it: in late 2025, IX.br announced an upgrade to 800 Gbps ports scheduled expressly ahead of the traffic peak it expected the World Cup to produce. The forecast proved right. During Brazil’s opening matches in June, the Sao Paulo facility hosting an IX.br node registered peaks of nearly 10 Tbps, the highest volumes its operator, Equinix, had ever recorded in the city.
In Europe, the pattern is measured in falling records. During the tournament, DE-CIX’s global interconnection ecosystem broke its all-time traffic record three times in under a month: the last mark, 29.5 Tbps, came during the Argentina vs Egypt round of 16 match, just one day after Portugal vs Spain had set the previous one. Italy, meanwhile, shows what happens when an entire football league migrates to streaming: since Serie A moved online, demand during major matches spikes 10 to 15 times above normal levels, and Namex, the Rome exchange, has seen its all-time peak more than triple in five years, crossing 1.4 Tbps during a Champions League night in January 2026.
Asia recorded its own inflection point a tournament earlier. When Japan’s ABEMA platform streamed every Qatar 2022 match free of charge, drawing a record 34 million weekly viewers, traffic across the country’s three major exchanges, JPIX, JPNAP, and BBIX, surged in a way no previous, television-only World Cup had produced. The lesson was unambiguous: the moment a tournament becomes an internet event, the exchange layer feels it first.
When “local” crosses a border
European exchange operators condensed all of this into a principle years ago: local traffic should stay local. Keep the exchange close to the audience, and live streams arrive with less latency, fewer detours, and more headroom when the peak hits. Almost all of that thinking, however, assumes that “local” is a national concept. One audience, one country, one exchange.
The 2026 World Cup, the first hosted by three neighboring countries, quietly stress-tested that assumption. Consider the Mexico vs England round of 16 match: with 13 million average streaming viewers across Peacock and Telemundo platforms, it became the most streamed soccer match in Spanish-language media history. Those streams originated on platforms based in the United States and were watched simultaneously by audiences on both sides of the US-Mexico border. For a few hours, “local” was binational.
That is not a one-off produced by a tournament. It is the everyday geography of this region: shared languages, shared platforms, shared fandoms, and audiences whose traffic naturally flows across the border in both directions. When that traffic has no nearby point to change hands, it detours through hubs hundreds of miles away, adding latency to precisely the kind of content that tolerates it least. Nobody wants to learn about a goal from a neighbor’s shout before it reaches their own screen.
The exchanges operating along the US-Mexico border, such as MEX-IX in McAllen and El Paso, exist to apply the European lesson to this binational reality: exchange the traffic where the audiences actually live.
The tournament ends. The pattern stays.
The final has been played, the records are in the books, and the traffic graphs will flatten until the next season begins. But the structural shift the World Cup made visible is not going anywhere. Italy’s arc shows the trajectory every market follows once live sport moves to streaming: peaks stop being anomalies and become the design baseline. Japan compressed that transition into a single tournament. Brazil is now engineering for it years in advance.
Live sport will keep migrating to streaming, league by league and market by market. Peaks will keep growing. And regions whose audiences are binational will need what every mature streaming market has already built: interconnection located where the audience is, not hundreds of miles from it. The World Cup did not create that requirement. It just made it impossible to ignore.
