The streaming industry spent its first era solving access: consumers no longer had to wait for a scheduled broadcast, buy a physical copy, or accept whatever happened to be showing. Streaming made enormous libraries available instantly across multiple screens and, increasingly, at multiple price points.
Then came a battle for content. Studios and technology companies spent billions assembling catalogues, acquiring franchises, commissioning originals, and securing sports rights. Content that was expected to attract subscribers, reduce churn, and differentiate one service from another.
But during my time at Cannes Lions last month, it became apparent that the industry’s next source of strategic advantage may lie elsewhere.
Today, control is emerging in the space between a viewer deciding that they want to watch something and actually pressing play: discovery.
The strategic question then extends from “Who owns the content?” to “Who understands what the viewer wants, identifies the right piece of content, and controls the route to playback?”
With this mindset, AI is no longer merely an improved recommendation feature for streaming. Instead, AI-powered discovery becomes a new distribution layer itself across streaming services, television operating systems, device manufacturers, and social platforms. And the competitive advantage at stake is no longer the content library. It is ownership of viewer intent.
Streaming solved access but created decision friction
Gracenote’s 2025 State of Play research found that viewers spent an average of 14 minutes searching for something to watch and almost 1 in 5 abandoned a viewing session when they couldn’t find suitable content.
The industry has created extraordinary content abundance, but that abundance has led to its own structural problem. A platform may have excellent options, but those options have little economic value if the audience cannot locate, understand relevance, or feel confident enough to select the content itself.
Entertainment time is finite. Deloitte’s Digital Media Trends research estimates that the average US consumer spends around six hours a day on media and entertainment – and streaming is competing for those hours not only with other services, but with YouTube, TikTok, gaming, podcasts, music, creators and a growing range of other digital entertainment options.
Social platforms in particular have trained audiences to expect immediate relevance, continuously interpreting behaviour to serve the next piece of content and removing the friction of a viewer manually navigating organised catalogues of content.
Streaming still asks viewers to do that work – to select an application, scan rows of options, and commit to a title. The gap between those two experiences is becoming increasingly visible.
From catalogue to conversation
But conversational AI today introduces a new form of discovery.
A viewer may not know the title they want. They may instead know they want something funny but not childish, a tense thriller that isn’t violent, or a story similar to a half-remembered scene.
These are not conventional catalogue queries. They are statements of intent.
Rather than retrieving what the viewer can name, an AI discovery system attempts to interpret what the viewer means, a distinction pivotal to the entire interface, where the application becomes the destination rather than necessarily remaining the starting point.
Every layer of the television stack wants to own discovery
This emerging battle is happening simultaneously across the entire television stack.
Netflix is exploring generative-AI-powered search that allows members to describe what they want in natural language, while developing recommendation models designed to predict the intention behind a particular viewing session, not only long-term user preferences.
Simultaneously, television operating systems sit in a particularly powerful position. Google has brought Gemini to Google TV; Amazon’s Fire TV supports natural-language discovery across topics, genres, plots, and remembered quotations; and Roku has introduced AI-powered recommendations across its home screen. These platforms are able to recommend across multiple services rather than only within a single catalogue, meaning they can influence which application receives the viewing session before the viewer enters any individual service.
Tubi’s launch of a native application within ChatGPT demonstrates a further possibility, where discovery can begin entirely outside the traditional streaming application. A general-purpose AI environment can sit above individual media applications and influence where an audience goes next, creating its own opportunity and risk for services acquiring new audiences but surrendering a portion of the discovery relationship to external platforms.
The next battle is the moment before play
Premium storytelling, live sports, trusted news, distinctive creators and powerful IP; these are all factors remaining fundamental to strategic advantage in television and are here to stay.
But content ownership alone may no longer guarantee control of distribution. As discovery becomes conversational and increasingly moves across applications, devices, and AI systems, the industry’s new strategic contest will centre on the moment between intention and playback.
The companies that understand viewers’ needs, identify the most relevant content, and control the viewing path will be the ones that influence attention, data, advertising, and commerce.
In an age of infinite choice, it is the platform that creates certainty that becomes more powerful than the platform that simply creates more options.
As we continue with this series, I’ll explore what this shift demands of the infrastructure that exists beneath it – from metadata and rights intelligence to the governance frameworks that will determine whether AI-powered discovery can be genuinely trusted at scale.
