Live, But Not Quite: Why Streaming Live Events Still Trail Broadcast
The big game is about to start, and you’re ready. Not long ago, that meant tuning your TV to the game through an antenna, cable or satellite box. Today, you can open a streaming app on a TV, laptop or phone, wherever you happen to be. It’s a huge convenience.
Then, in the middle of the game, the neighbors cheer through the wall, or your phone buzzes with a score alert, while nothing has happened on your screen yet. Are you really watching what is happening right now, or what happened a minute ago? A minute is no exaggeration. At the last two Super Bowls, measurements by latency specialist Phenix in 2025 [1] and by Stats Perform in 2026 [2] put average streaming delays anywhere from 26 to 78 seconds behind the action.
That isn’t exactly watching the game “live,” is it? And there are specific reasons for it.
More than a decade ago, as a Solution Architect at Ericsson, I helped design and deliver an end-to-end direct-to-home (DTH) satellite TV system based on DVB-S2. Today, streaming is how millions of people watch live sports. The shift is remarkable, but it came with certain trade-offs.
It’s not your router
When experiencing an issue with a live stream, many viewers assume the problem is on their end. They see a spinning wheel and blame the network they know and pay for: their own. It is the same instinct that made many viewers think their cable had failed when The Sopranos abruptly cut to black [3]. In a recent survey, 31% of NFL streamers said they had upgraded their internet plan, Wi-Fi or router to stream sports better, and another 31% had considered it [4]. While a better connection can reduce stalls, it does little to address the delay introduced by encoding, packaging and buffering throughout the delivery chain.
Built for on-demand
Much of today’s HTTP-based streaming relies on techniques well suited to video on demand (VoD), where an uninterrupted, good-quality picture matters far more than timing. If a movie player buffers 30 seconds of video, nobody notices. For a live game, that same cushion adds 30 seconds to the delay behind the action.
Video is divided into short segments that ordinary web servers can deliver. Content delivery networks (CDNs) cache popular titles near viewers, while the player buffers video segments against interruptions and adjusts picture quality when bandwidth drops. Live streaming uses much of the same architecture, but buffering introduces a trade-off: reliability at the cost of added delay.
How the seconds add up
Live content passes through two broad stages before it reaches the viewer. First comes the network, or content provider, such as ABC or Fox, which shoots and produces the game: cameras, shot selection, graphics, master control, and, in some cases, an intentional broadcast delay [5]. The network then hands its feed to local stations, cable and satellite operators, and streaming services, using satellite, fiber or IP connections.
Depending on the system, distributors may decode, re-encode, encryptm or repackage the feed, adding further delay. At Super Bowl LX, Stats Perform measured over-the-air TV at 19 seconds behind the field and cable at 38, compared with 48 seconds for Peacock and 53 to 62 for other streaming services [6]. In both of the last two Super Bowls, the network’s own service was the fastest stream: Fox’s Tubi in 2025 and NBC’s Peacock in 2026 [7]. This suggests that a more direct feed can avoid additional processing steps, and some of the delay that comes with them.
The second stage, distribution to the viewer, is where the technologies really differ. In the DTH system I worked on, that stage ran from signal acquisition all the way to the subscriber’s set-top box. Channels were encoded, multiplexed, scrambled and transmitted to a geostationary satellite 22,300 miles up and back down to the dish, a journey that, despite the enormous distance, takes only about a quarter of a second. Because the signal arrives at a constant rate, the set-top box needs only a small buffer. In that system, this stage added about 4 to 6 seconds, mostly from encoding and buffering. This excludes earlier production delays. The added delay was predictable and consistent for viewers receiving the same feed.
Streaming’s distribution stage has its own specifics. On the streaming platform, video is encoded at multiple bitrates, packaged into segments typically lasting 2 to 6 seconds, and sent through origin servers, CDNs, the viewer’s ISP and home network. With conventional streaming, the player receives complete segments and buffers additional video before playback. Satellite’s distribution stage adds a few fixed seconds; conventional streaming generally adds more, and part of that delay varies by viewer. It depends on network conditions and load, device behavior and how the player responds. A stall can push a viewer further behind live. At Super Bowl LIX, Phenix found that the gap between the fastest and slowest viewers of a single service ranged from 39 to 57 seconds [8].
Who controls the path
Satellite and traditional cable operators largely control their distribution networks. Streaming services generally don’t control the entire route. Their video may pass through CDN infrastructure, an interconnection with the viewer’s ISP, the ISP’s backbone and metro network, and finally a shared access network, such as a cable node or a fiber split serving dozens of homes. Some over-the-top (OTT) services run their own CDNs, private backbones or caches inside ISPs, but most cannot control the viewer’s ISP network or the shared access link. Networks are engineered for busy-hour traffic, but a major live event can still create bottlenecks when millions tune in at once. Once a home connection is fast enough to carry the stream, a faster plan raises only that viewer’s speed limit, not the capacity of the shared links upstream. The player’s buffering is insurance against that uncertainty, and the insurance is the delay.
CDNs remain essential, but a live segment can’t be cached near viewers in advance: it is created only seconds before millions request it at once. And unlike VoD, a live player cannot download video that has not yet been produced. To build a larger buffer, it must play farther behind the action: every second of protection is another second of delay.
Then there is scale. A broadcast signal needs the same capacity whether one viewer tunes in or ten million. Streaming needs a separate delivery connection for each viewer, even when CDN caches serve the same segments to many.
Closing the gap
The most direct fix is to reduce packaging and buffering delays. Chunked CMAF with LL-HLS or LL-DASH delivers video in small pieces before complete segments are ready, substantially cutting delivery delay. The cost is a thinner buffer and more risk of stalls, which is why some services, YouTube TV among them, offer low latency as an option [9]. Real-time protocols such as WebRTC can go below a second, but are harder and costlier to scale to millions of viewers.
Another is to take more control of delivery: more direct program feeds, caching inside carriers’ networks and multicast where supported, so one transmission can serve multiple viewers.
Finally, streaming can borrow from broadcast. ATSC 3.0, the next-generation over-the-air TV standard, lets a TV combine a broadcast signal with an internet connection, and 5G Broadcast brings broadcast reception to compatible phones [10]. In both, one transmission serves every viewer in range, removing a major source of congestion and variable delay, although encoding, packaging and receiver buffering still add latency. These approaches use broadcast delivery rather than conventional per-viewer OTT streaming: they need spectrum, transmitters and compatible receivers that OTT services don’t control, which makes them most attractive to broadcasters with their own streaming services, or to OTT providers that partner with them.
For viewers, choosing a lower-delay setting where available, or restarting a stream that has fallen behind, can help. Neither, however, eliminates the delay built into the service.
So has streaming made live TV better? More accessible and flexible, yes. More “live”? Not yet.
The gap reflects engineering choices about reliability, scale, and distribution. Streaming is where live TV is heading, and in time it will likely close much of this gap itself. Until then, it can get closer to live by reducing unnecessary processing and buffering, and by borrowing more from broadcast rather than replacing it outright.
[Editor's note: This is a contributed article from Quandary Park Research. Streaming Media accepts vendor bylines based solely on their value to our readers.]
References
[1] Phenix, “Super Bowl Latency Continues in the Era of Social Media,” Phenix blog, February 2025, https://blog.phenixrts.com/phenix-superbowl-latency-study-2025; see also Phenix, “Latency comparison of common streaming services for the 2025 Super Bowl,” https://www.phenixrts.com/resource/super-bowl-2025
[2] Stats Perform, “2026 Super Bowl – Latency Study,” February 2026, https://www.statsperform.com/wp-content/uploads/2026/02/2026-Super-Bowl-%E2%80%93-Latency-Study.pdf; per-service figures as reported by The Desk, February 2026, https://thedesk.net/2026/02/stats-perform-phenix-latency-super-bowl-lx/
[3] “When The Sopranos Blacked Out,” Mental Floss, https://www.mentalfloss.com/article/501524/10-years-later-when-sopranos-blacked-out; Steve Donohue, “Cable Ops: Viewers OK with Sopranos Finale,” Multichannel News, June 11, 2007, https://www.nexttv.com/news/cable-ops-viewers-ok-sopranos-finale-335808
[4] Chris Pugh, “Buffering, freezing top NFL fans’ streaming frustrations,” MassLive via Yahoo Sports, October 21, 2026, https://sports.yahoo.com/articles/buffering-freezing-top-nfl-fans-111202887.html
[5] For example, NBC’s Super Bowl XLVI telecast ran on a five-second delay: “NBC Missed Blurring M.I.A. Obscene Gesture by ? of a Second,” Sports Video Group, February 7, 2012, https://www.sportsvideo.org/2012/02/07/nbc-missed-blurring-m-i-a-obscene-gesture-by-2-3-of-a-second/; Fox used a five-second delay for its Super Bowl LI pregame and halftime shows: “Fox to use five-second delay for pregame, halftime shows,” ProFootballTalk, February 5, 2017, https://www.nbcsports.com/nfl/profootballtalk/rumor-mill/news/fox-to-use-five-second-delay-for-pregame-halftime-shows
[6] Stats Perform, “2026 Super Bowl – Latency Study” [2]; per-service figures as reported by The Desk, February 2026, https://thedesk.net/2026/02/stats-perform-phenix-latency-super-bowl-lx/
[7] Phenix measured Tubi at 26 seconds at Super Bowl LIX, the lowest of the seven services it tested: Phenix [1]; “Super Bowl LIX: Tubi Leads All Streaming Platforms in Least Lag Time, According to Phenix Report,” Sports Video Group, February 10, 2025, https://www.sportsvideo.org/2025/02/10/super-bowl-lix-tubi-leads-all-streaming-platforms-in-least-lag-time-according-to-phenix-report/. Peacock at 48 seconds versus 53 to 62 for other services at Super Bowl LX: see [6].
[8] Phenix defines drift as the difference between the slowest and fastest streams of a service; YouTube TV showed 57 seconds and Tubi 39 seconds. Phenix [1].
[9] For example, Comcast’s comparison for Super Bowl LX lists YouTube TV at 26 seconds with its low-latency option and 39 seconds on standard latency: “Comcast: Peacock Super Bowl stream averaged 17 seconds of latency on Xfinity,” NewscastStudio, February 20, 2026, https://www.newscaststudio.com/2026/02/20/xfinity-super-bowl-lx-latency/
[10] 5G-MAG, “5G Broadcast,” https://www.5g-mag.com/standards/5g-broadcast
Related Articles
ATSC, The Broadcast Standards Association®, today announced an important milestone in the development of Broadcast-to-Everything (B2X), with the approval of a new Candidate Standard that establishes a critical foundation for the emerging B2X platform.
08 Oct 2026
That gap between reality and your screen is latency, and it has become one of the most commercially consequential technical problems in the live sports industry. With American sports betting exploding in scale and prediction markets rewriting the rules of fan engagement, the stakes around streaming delay have jumped from a quality-of-experience nuisance to a structural business problem that touches broadcasters, technology companies, regulators, and sportsbooks alike.
10 Mar 2026
In this article, we'll look at the state of real-time streaming a few months into 2023, including several use cases that hold longer-term promises. But before we do so, here's a baseline definition of real-time streaming, as used in the survey: Real-time streaming is "device-synchronized delivery to hundreds of thousands of viewers at less than 500 milliseconds per user."
12 Apr 2023