Big Blue Marble says it delivered more than 10 terabits per second of streaming traffic through Amazon CloudFront during a high-demand World Cup match for a European public-service broadcaster.
The figure is striking, but it is a measurement of bandwidth—not a disclosed audience count and not, by itself, evidence that every viewer received a reliable stream.
The useful engineering story begins with the questions behind that number.
Why 10 Tbps does not equal 10 million viewers
A delivery peak depends on the number of concurrent streams and the bitrate delivered to each device. One million viewers receiving 10 megabits per second would account for roughly 10 terabits per second before overhead; a larger audience at a lower average bitrate or a smaller audience receiving high-bitrate UHD could produce a similar total.
Adaptive streaming complicates the calculation further. Different viewers may move among several quality levels as their device, connection and player buffer change. Replays, second-screen clips, advertising and authentication requests can add load that is not represented by one simple audience number.
The edge network carries the emotional peak
Live sports traffic is unusually volatile. A goal, penalty or disputed call can cause large numbers of viewers to return simultaneously, request replays or open additional streams. The delivery system must absorb those surges before players run out of buffered video.
Amazon reported that CloudFront carried more than 117 Tbps across more than 40 broadcasters at the peak of the 2026 World Cup final. It says embedded points of presence inside internet-service-provider networks, origin protection, traffic rebalancing and rapid failover helped keep segments close to viewers and reduce pressure on the originating systems.
That larger AWS total is not directly comparable to Big Blue Marble’s single-broadcaster report, but it gives useful scale and architectural context.
The measurements broadcasters should demand
- Peak and average concurrent viewers, separated by device and region.
- Average delivered bitrate and the share of viewing in HD and UHD.
- Startup time, rebuffering rate, fatal playback errors and latency behind the live event.
- Cache-hit ratio and the traffic returned to the broadcaster’s origin.
- Capacity and performance of backup content-delivery networks.
- Time to detect, route around and recover from an edge, origin or third-party failure.
Broadcasters also need to understand where responsibility changes hands. The production may be clean while distribution fails; the content-delivery network may be healthy while authentication or advertising breaks playback. Monitoring must follow the complete viewer path rather than stopping at the encoder or origin.
The 10 Tbps report is valuable because it illustrates the scale of one national live event. Its real usefulness will come when Big Blue Marble and the broadcaster disclose the accompanying quality and audience measurements. High throughput is impressive. A stream that starts quickly, remains synchronized and survives the biggest moment is the result that viewers actually experience.