AI Content Deals Hit New High
Microsoft, Meta, OpenAI, and Disney are all cutting checks. From news to music to movies, here’s what’s driving the record-breaking surge in AI licensing deals.
AI content licensing deals are spiking—and fast.
I’ve been tracking every publicly announced deal since 2023, and the numbers are clear: we’re now seeing more AI-content partnerships than ever before.
Here’s the volume by month:
What’s behind the surge?
I see four forces at play:
1) Bot Blocking — Cloudflare’s Power to the People
AI models need content.
So when Cloudflare launched its AI bot blocking features in July 2025, it put AI models on notice (Cloudflare represents 20% of websites).
There were 10 reported content licensing deals before Cloudflare’s bot blocking.
There have been 22 since.
Many AI companies would rather pay for content than lose access to valuable content.
2) Microsoft Needs a Backup to OpenAI
It should be no surprise that Microsoft is now striking content licensing deals (Gannett, People Inc, Condé Nast).
Microsoft’s OpenAI Partnership (which Sam Altman once called “The best bromance in tech”) is now less romance and more risk management.
OpenAI now gets compute from Microsoft rivals AWS and Google Cloud.
And Microsoft invested in OpenAI rivals Anthropic and Le Mistral.
The new content Microsoft is licensing is for its Publisher Content Marketplace will initially be used for Microsoft’s Copilot (based mostly on ChatGPT).
But I expect Microsoft to also pay publishers to use content on its own LLM (called MAI (“Microsoft artificial intelligence”)) too.
3) Meta Plays Catch-Up with “Frontier Four”
This month, Meta signed AI content licensing deals with CNN, Fox News, Le Monde Group, People Inc., and USA TODAY Co., Inc.
The last announced content deal they did before this was with Reuters in October of 2024.
My take is that Zuck just wants to cross every item he can off his list to have an LLM that can compete with “The Frontier Four” (OpenAI, Anthropic, xAI and Google).
4) A Demand for Multimodal Content
The first AI content licensing deals were mostly text-based with some images.
Now, you’ve got sound and video.
Last month, Universal, Sony and Warner all signed licensing deals with AI music model Klay.
Warner signed another deal with Suno.
And OpenAI and Disney signed their $1B deal — arguably the biggest AI licensing deal.
And the mega-deal is only for just 50 of Disney’s characters’ use in just one of OpenAI’s products (Sora).
AI models need more than just text — they need audio, video, and images.
And I expect this trend to continue as new AI models come out — like I wrote about last week in “The 7 Content Types Most Valuable to AI World Models”.
The Impact from Court Cases
Of course, the number of AI content licensing deals in the future will be heavily impacted by the outcomes of these active cases:
NYT v. OpenAI/Microsoft — flagship text training, pricing leverage.
Getty v. Stability AI — clean image facts, dataset licensing pressure.
Dow Jones/NY Post v. Perplexity — RAG precedent, AI search monetization.
Ziff Davis v. OpenAI — important, narrower publisher portfolio.
Authors v. Meta — fair-use limits, weaker licensing leverage.
Disney v. Midjourney — massive optics, slower licensing signal.
Thanks for reading!
Rob Kelly, Creator & Host of Media & the Machine
p.s. How to reach me? —The best way is to subscribe below and reply to any of my emails. There’s a free option — and your reply comes directly to my inbox.




Hey, hey, Rob!
One thing that stands out to me across all four forces is how quickly content licensing is shifting from optional to infrastructural.
Bot blocking, competitive hedging, multimodal demand, and legal pressure are all pointing to the same underlying reality. Access to content is no longer a growth accelerator. It is a prerequisite for staying in the game.
What also feels different this time is who holds the leverage. It is not just the biggest publishers. It is the ones with rights clarity, provenance, and content that is actually usable across training, retrieval, and multimodal workflows. In that sense, the court cases almost matter less than the operational readiness of the content owners themselves.
It feels like we are moving into a phase where AI companies will care less about scraping more and more data and more about securing fewer, cleaner, defensible inputs. That changes how deals get done and who wins them.
Curious if you are seeing early signals of AI companies prioritizing content that is not just licensable, but structurally prepared for different model types and use cases. That feels like the next layer of differentiation.
Keep up the great Media and Machine research!
Dan