When AI meets live sports
Lights, camera, competition.
An India-Pakistan match brought me back to watching Cricket last year. I followed the match on Cricinfo until I got to my hotel in Taipei, and then thought I’d try my luck with TV: Hotstar doesn’t allow paying subscribers to stream a match outside of India. A hundred channels on TV and not one with the match. Then I did what any normal person who wants to watch a match he can, does: search YouTube.
On searching YouTube, I found something less sufficient, but much more interesting: cricket commentary channels.
The commentators were exceptional: emotive, with detailed situational analysis of the proceedings and the state of the match, not just of every ball bowled and shot played. They had knowledge of field placements, and player statistics. It was live, continuous, and descriptive: the next best thing to watching it play out in front of your eyes.
They kept switching between commentators every 45 minutes, and each commentator brought a fresh analysis to the proceedings, not just a fresh voice. This was YouTube, and on the screen they had statistical analysis and data, that kept changing with each delivery. Every few balls, when a boundary was hit, a part of the screen showcased an animation of the action. Similarly, there were animations depicting wickets and catches.
They got about 5.5 million streams during the match, and this wasn’t the only YouTube channel with live match commentary.
Vibe coding with AI - Workshop with Nikhil Pahwa
Build a real app with AI, your personal website, a mini game.
Everyone who attended our last workshop went back having built at least three things.
None of them knew coding coming into the workshop. They still don’t, but they can build with AI.
One day. Max 15 people. Hands-on from start to finish.
Who’s running it:
→ Nikhil Pahwa, Founder of MediaNama and Reasoned.live — who has vibe coded 20+ real products himself (some at: upthink.app)
📅 Friday, 5 June 2026 | Full Day
📍 Central Delhi
🎟️ ₹15,000 + taxes
(P.s.: if you can’t make it, tell a friend)
What Sports Broadcasting Tried With Hot News
Back in 2013, STAR TV took Cricbuzz founder Piyush Agarwal, Onmobile Global and Idea Cellular to court over the distribution of live cricket scores. Since they couldn’t claim copyright over facts, STAR argued that BCCI held “Quasi Property Rights” over Cricket scores because the events were organised by them. STAR had paid Rs. 3851 crores for these rights across TV, audio, radio, Internet and mobile. From the article I wrote then:
Singhvi called Idea’s act of sending scores without sharing revenue with STAR (and hence the BCCI), an unlawful appropriation of property. The information is time sensitive, Singhvi said, saying that the score is relevant for 5-10 min, and it has all the attributes of a commercial property. “Time sensitive information has the capability of becoming commercially viable for exploitation,” he added. He emphasised that this case is between an owner of the property, and someone using it for commercial exploitation. STAR, he said, has no issues when a user sends a score to another for non commercial purposes. The matter is one of transferring information for money, in competition with STAR.
This was called Hot News. Before courts intervened, commentary on websites and SMS was “deferred live” — delayed by 5 minutes. The case has been in limbo for more than a decade; the Supreme Court of India is yet to hear it, as far as I know.
The importance of live to streaming
Netflix, Amazon Prime, Sony Liv and Hotstar brought time-shifted television to India, shifting behaviour away from appointment viewing. Without live events, it becomes almost impossible to predict when large audiences will gather — and that’s exactly when brands get maximum value. Live sports remains the exception. It’s why Star, under James Murdoch and Uday Shankar, bet big not just on acquiring streaming rights but on setting up leagues — manufacturing enough live “events” to attract advertisers. Cricket matches draw millions; advertisers build campaigns around specific tournaments the way the SuperBowl does in the US.
Live streaming is what makes streaming services most of their money: it’s the last construct of scarcity in the abundance of on-demand content.
Not all doom and gloom
At the recently concluded AI Summit, Reliance Jio demonstrated live translation of commentary, deferred only by 2-4 seconds, they claimed. I heard an audio of Stuart Broad (the Cricketer) speaking in Hindi. The person at the booth said (I’m translating and paraphrasing a little):
“We’ve built a software layer where I speak in English and the output comes in Hindi, or any other language, in my own voice. It’s voice-to-voice. You don’t hear a different commentator. It still sounds like me.”
“Right now the delay is about two to four seconds. The system needs to detect when a sentence has completed before translating it. Two seconds is actually very fast for that. Even some of the big tech teams were surprised it could be done this quickly.”
“Imagine a situation where the Hindi commentary box wants insights from someone like Stuart Broad. He can’t speak Hindi. But he can speak in English, and the output comes instantly in Hindi. You break the language barrier without losing the insight.”
This uses AI to decouple insight from language, and makes commentary portable to different languages. At that time, they demo’ed about 6 different languages. What was truly remarkable was that it didn’t take them long to build this:
“To be honest, this was built just last week for the AI Summit expo. It’s actually very easy to build now. There are open-source tools, APIs from companies like Google or AWS. You just assemble them into a single interface. The real work is in filtering names and improving accuracy.”
So how can AI impact live sports broadcasting?
First, AI makes commentary cheap: A YouTube channel translating TV commentary costs far less than hiring commentators in multiple languages. This does not replace the broadcast, but it does act as an imperfect substitute. The YouTube commentary I experienced was far superior to text-based coverage on Cricinfo. If audio is protected by copyright, you will probably be able to create your own by pointing a camera at the match and letting a trained LLM do the rest. The commentary can be generated from the state of play: field placements, score, player histories, match context, and visual cues.
Second, AI can eventually generate match visuals: If a channel can infer what is happening from a live feed, score data, or even partial visual input, it can produce an animated version of the match. Not the match itself, but a live reconstruction of it. It isn’t the same as, as Singhvi had mentioned earlier, a telescopic camera on the rooftop outside the ground. For many viewers, especially those who are blocked from the official stream, that may be enough. It’s not like there isn’t animation already in live sports: Venkat Ananth pointed out to me that F1 Kids had 3D avatars for drivers to try and attract kids to Formula 1 (an example here), but the race is very real.
This creates a problem that copyright law is poorly equipped to handle, and perhaps a revival of the Hot News debate is pending. The substitute can now replace experience of watching the broadcast live, though not exactly an exact substitute.
More than that, sports broadcasters will have to become a little more customer centric. In Taipei, I just wanted to watch a match I was already emotionally invested in. Music has shown us this: people are willing to pay, but pirate when restricted by geography, exorbitant pricing or platform boundaries. Broadcasters need to do better. Sometimes an imperfect substitute is enough.
Found this interesting? Share it with someone who might find it useful.