Video is the dominant medium of our time — for news, commerce, sports, education and entertainment. And while AI is the media industry’s greatest opportunity, it’s also its most pressing challenge.
Despite all the AI hype, only 14% of M&E firms report being ready to handle AI-driven content creation and oversight, largely due to disjointed processes and incomplete metadata, reports Ernst & Young. Unfortunately, most media companies still can't answer the most basic questions about their own content: What do they own? Where is it? Can they legally use it?
The answer, almost universally, is: they don't know.
Media organizations are sitting on terabytes (and often petabytes) of untapped intellectual property. Decades of video, audio, scripts, stills, and metadata lie scattered across aging asset management systems, on-prem archives, cloud buckets, and third-party platforms. Most of it is effectively dark: inconsistently tagged, siloed by business unit, impossible to search.
"The biggest issue by far is that most media organizations do not have control of their data and their IP to make it usable at the speed and scale that AI requires," says Simon Crownshaw, head of M&E at Microsoft. "Ultimately, this is a data problem; this is a content storage problem."
The organizations that solve it will dominate the AI era. Here are the six capabilities they need.
1. Ingest and store with provenance
Content lives across tapes, disk arrays, multiple clouds, and partner systems, often with murky ownership and incomplete rights history. An AI-ready platform tracks what each asset is, how it was created, which versions exist, who can access it, and which restrictions apply. This isn't a nice-to-have. Without a single governed source of truth, every AI initiative built on top is built on sand.
2. Understand and index semantically
Keyword search can find a title; it can't find the shot. Editors and journalists need to find specific scenes, faces, emotional beats, and topics across years of footage, not just clips that match a title. AI-enabled platforms analyze video, audio, and text together, building semantic indexes at the shot level.
A breaking news team can now query: "late-night protest clips in the rain with our main correspondent and visible police presence." What used to take an hour takes seconds.
New York public media giant WNET Group proved this at scale. Working with Microsoft, the PBS parent organization migrated 3.6 petabytes of content to the cloud, while maintaining a 24/7 broadcast operation — a feat its director of technology solutions Ryan Weston calls "game changing." The migration unlocked the ability to run machine learning across WNET's entire archive. Producers can retrieve archival footage the same day, a gain that delivers greater agility in breaking news situations and in the longer term, it could facilitate new workflows and content reuse.
As a station that’s aways on air, “this bus is never stopping, but we are still changing the wheels,” explains Weston.
3. Build a knowledge graph of stories, characters, and rights
Knowledge about franchises, characters, storylines, and rights windows is trapped in spreadsheets, legal documents, and the memories of people who may not be there next year. A content intelligence platform builds a knowledge graph connecting all of it — characters, storylines, locations, rights windows, and performance data — so teams across legal, marketing, and editorial share a common map. For news organizations, it connects topics, events, sources, and archive packages into coherent topic universes. Which storylines are safe to license in a given territory next quarter? Which climate explainers have evergreen value worth recutting? The graph answers these questions in real time.
4. Creative generation grounded in the catalog
Generic AI models that don't understand a company's canon, brand, or rights can't be trusted in production. A content intelligence platform uses the semantically indexed catalog as the grounding layer, generating briefs, storylines, mood boards, and promos based only on approved, rights-cleared material.
For local newsrooms, the results can be transformative. Nota, an AI-enabled SaaS startup, used Microsoft Azure's OpenAI service to build tools that turn written articles into engaging videos — tools that take a journalist about 10 minutes to master and begin optimizing articles and creating videos. Nota reports 10 times better engagement and revenue than text-only stories. Videos are tagged and structured to optimize organic traffic.
“That’s very important, because small newsrooms can live and die by how quickly they are able to break stories,” says Josh Brandau, Nota’s CEO. “Microsoft has shown us that we don't have to be a gigantic organization with a huge R&D lab to make exciting AI applications. It gave us the ability to achieve tremendous progress in a very short period of time."
5. Pattern recognition and prediction
Ratings and click data tell what happened. AI tells why and what to do next. Using semantic data and the knowledge graph, AI can surface patterns in formats, talent combinations, story arcs, and release strategies, then flag novel combinations worth testing. It might reveal that a specific character dynamic performs well with certain topics in particular regions, or that particular explainer formats have long-tail value when repackaged.
Consider a major entertainment studio sitting on decades of franchise IP. AI pattern recognition can analyze which character pairings, narrative arcs, and release windows have historically driven the strongest performance — by region, platform, and audience segment — then surface untested combinations worth greenlighting. A sequel greenlit on gut instinct becomes a decision grounded in catalog intelligence.
"The future of news, sports, episodic content, and movies is the IP and data behind what you already have," says Crownshaw. "If you can't enable that, you are in big trouble."
6. Governance and security for IP and trust
As synthetic media proliferates, governance has become as critical as capability. For Microsoft's Crownshaw, it is less about regulatory compliance than protecting the asset that will power the next decade of growth.
Many of the risky, short-term licensing deals being struck today are, in Crownshaw's view, fundamentally governance failures: companies making decisions without fully understanding what they're surrendering. Governance also shapes the AI ecosystem choices organizations make — which models they use, where those models run, and who controls their evolution. Crownshaw argues the industry will shift toward open-source models precisely because they reduce legal exposure and allow companies to build on top of their own IP rather than ceding it to opaque third-party platforms.
Nowhere is this more acute than in news, where trust, provenance, and speed collide. Newsrooms must ingest third-party and user-generated content at high velocity while upholding strict standards around authentication and content security. For Microsoft, enabling governance means giving media organizations the architecture to retain ownership of their IP and signals, govern their AI model choices, and validate content at speed, without surrendering what makes them competitively unique.
"IP is the only thing that differentiates you down the road," says Crownshaw. "It is your lifeblood. Never give it up, ever."
Start in 90 days, not three years
The answer isn't another massive multi-year data transformation program. It's a focused 90-day sprint to prove value fast.
Pick a pilot corpus — one franchise, one news vertical, a decade of flagship programming. Choose a slice with clear ownership and obvious pain points. Define hard metrics: speed-to-discovery, enrichment accuracy, turnaround time, catalog reuse. Then stand up the core capabilities for that slice: ingest with provenance, semantic indexing, a lean knowledge graph, and one or two AI-assisted workflows. Standardize the schemas and pipelines so the next corpus scales faster.
The goal isn't modern storage. It's a content intelligence platform where every journalist, producer, and marketer can ask questions of the archive and get trusted, actionable answers in seconds.
The window to act is open, but it won't stay that way. Organizations that start small, prove fast, and scale smart won't just modernize their operations. They'll compound that advantage with every sprint that follows.
To learn more how Microsoft is helping media leaders turn AI into measurable impact, visit here.