Notes from IBC 2026: linear television is becoming one output among several, which quietly changes what broadcasters have to plan. AI, for all the noise around it, makes the change more urgent.
The sharpest line I heard at IBC this year was four words long. Someone suggested that 83 % of broadcasters have not yet integrated AI into their core processes. Someone else replied: “And 17 % are lying.”
It was a joke about AI, but also about accounting – about how hard it has become to say what is actually happening inside a broadcast operation. I work for a company whose first broadcast product was a linear scheduler, and that has left me with an occupational hazard: show me anything and I want to know where it runs, when, and who pays for it. Ask those questions of IBC 2026 and the AI noise gives way to a quieter story. Linear television is becoming one output among several, and the unit we plan is shifting from the slot to the content item.
A live show no longer ends when the transmission does. Teasers come before it and highlights after, keeping the audience warm between broadcasts; in sport, the behind-the-scenes footage may be the most valuable thing a club has between matches. The industry has been saying this at IBC for years. Two things this year suggest it has stopped being a forecast.
In September, BBC Studios announced two new online Top Gear channels that launch before the programme returns to BBC One after four years off air. Online goes first; linear becomes the culmination of the build-up rather than the source that gets chopped into clips. And in June, the Reuters Institute’s Digital News Report found that, for the first time, social media and video platforms had overtaken TV as a source of news globally.
Every door is also an exit
A platform representative at IBC explained that unless broadcasters integrate fully with the platform, their viewers will drift there anyway – and stay for someone else’s content. Nice audience you have there; shame if anything happened to it. The remedy on offer was, naturally, more of the platform.
The irritating part is that the diagnosis is right. A viewer follows a clip, then another, and never comes back. No broadcaster controls someone else’s recommendation engine. What a broadcaster can control is what each piece of content is for: which programme it points to, when it goes out, what it is allowed to show and what it costs. That, unglamorously, is planning.
The promo that left the building
At many of our customers, that planning happens somewhere else. Digital and social teams plan clips and edits in their own tools, spreadsheets included, and the linear schedule never sees them. Ask my three questions of a single reel and you may get three answers from three places – and none from the schedule.
Schedulers have treated promos as given items for decades. Much of today’s social output is the promo that left the building – and the schedule never followed it. That is what the shift to the content item means: derivatives – the clip, the edit, the reel – become planned items in their own right, linked to the programme they serve, with a cost, a rights position and an outcome. Not every one can be planned by hand; some are made in minutes, some by machines. The realistic goal is visibility: rules, budgets and rights for whole families of content.
Linear is not going anywhere. It remains the anchor and the till. But its value increasingly depends on what surrounds it – and whether all that surrounding effort pays, or mostly heats someone else’s house, is hard to tell when the costs sit in one spreadsheet and the results on a platform’s dashboard.
The boring bits are the good bits
AI makes all of this more urgent. Everybody at IBC had an AI, and some stands seemed to have more agents than visitors. Thousands of AI-generated or hybrid microdramas now appear every day. Efficiency, it turns out, brings its own inefficiency: the more content there is, the harder it is to find the piece that is genuinely better. Guidance and trust are becoming products in their own right – which makes a well-made schedule, the oldest form of curation in our business, worth more, not less, when content is this plentiful.
Which brings me back to the 17 %. The AI successes I actually believe in are rather boring: archive search, voice conversion and dubbing, news articles drafted from transcripts, and sports highlights published minutes after the final whistle without a human touching them. Those highlights turn one match into a stream of items. The AI that reliably works today is, among other things, a machine for multiplying exactly the content our schedules cannot see.
People do not disappear from this picture. They move from making the clip to answering for it – for the rights and for the judgement calls. Seniority therefore counts for more, and the agents themselves will need something like an HR department for bias and security. I await the first agent performance review with interest.
An AI filmmaker at IBC compared this moment to the arrival of sound in the 1920s – not because the early talkies were better films, but because they widened what a film could be. Early results today are often underwhelming and costly, and some paths will lead nowhere. Still, I am absolutely persuaded there is light at the end of the tunnel. Nobody has the map yet.
A place in the plan
At PROVYS, linear is where we come from, and we have supported our customers’ non-linear operations for some 15 years. The direction we are working towards treats the programme, the edit and the reel as equal content items – each with metadata, a budget and a place in the plan, each aware of the others – building on the AI assistants and agent integration we already ship. It is a direction, not a finished product, and we are looking for a customer or two willing to go through the early stages with us. We won’t pretend to hold the map, but we can keep a light on.
I can’t promise we’ll find the way. I can promise that when we say we have, we won’t be in the 17 %.
Pavel Koběrský, Product director
To learn more about AI in Sphere, read AI agents join the team.
