BREAKING: Democrats win Senate control in midterm setback for Trump; House projected Democratic (AP call pending)
A vintage film camera on a studio soundstage facing a console that projects glowing holographic film frames through gold light filaments

Studios are taking equity in AI labs, streamers are building in-house tools, and agencies are buying their way into the stack. The gate is not gone. It is being redesigned.

Hollywood’s oldest power map ran through a handful of agencies: a script, a star, a package, a green light. That map is being redrawn, not torn up. US tech and film companies are building AI-native pipelines that let studios, streamers and social-first creators move from idea to image with fewer traditional middlemen. The agencies have noticed. They are trying to stay inside the new workflow, on their own terms.

The shift matters this week for another reason. As Nvidia and the chipmakers lost hundreds of billions in market cap last week, the companies spending those chips on storytelling kept cutting deals.

The studio that bought a seat at the model

The clearest studio-side move is Lionsgate and Runway. In June 2026 they expanded a partnership first struck in September 2024. Lionsgate took an equity interest in the generative video firm and announced a joint development programme for new intellectual property, starting with a short-form episodic series drawn from Lionsgate’s catalogue and Runway’s models. Runway said so in its own announcement. Variety reported the same deal, noted that financial terms were not disclosed, and said Lionsgate’s stake was not a cash investment.

The new architecture is simple: a legacy studio does not just license a tool. It takes a piece of the lab, then feeds its own IP through an AI pipeline. The 2024 phase was pre-visualization, storyboarding and parts of post-production. The 2026 phase is new shows. Lionsgate has also put a Chief AI Officer, Kathleen Grace, and an AI steering committee on the job.

AI-native studios and creators going direct

A parallel track is even more direct. Variety has described a wave of AI entertainment studios that design workflows around generative tools from day one. The list included Promise, the venture backed by Peter Chernin’s North Road and Andreessen Horowitz; Asteria; TCL Studios; and social-native outfits such as Toonstar and Invisible Universe, which make original IP for YouTube and other feeds. The pitch: smaller budgets, shorter timelines, more output.

Asteria is now part of that pipeline in a second way. Moonvalley, a Toronto-based model company, acquired the studio and later released Marey, a video model it says was trained on licensed data rather than scraped web video. The Ankler reported that CAA invested in Moonvalley’s $84 million Series B, taking total funding to $154 million, with Alexandra Shannon, CAA’s head of strategic business development, framing the bet around “consent and control” for talent.

Streamers are in the same race. Reuters reported in February 2026 that Amazon planned to use AI to speed up film and television production through an AI Studio initiative at Amazon MGM, led by Albert Cheng, with a closed beta for industry partners.

The social-first shops in Variety’s survey are the loudest version of going around the old door, building worlds for YouTube and TikTok-speed feeds where an agency package is optional. Runway has also been convening filmmakers at the Runway AI Festival, with Lionsgate as a presenting partner in June.

None of this means CAA, WME or UTA have been written out of town. Packaging a star or booking a private appearance still runs through the existing talent economy. Celebrityfees keeps live book-a-celebrity routes and a public read on the cost to book an A-lister for a private show. The shift is real. It is also incomplete.

How the agencies are answering

The agencies are splitting into different defensive shapes. TheWrap has reported that WME is leaning on technology partnerships rather than building its own digital vault, with a focus on partnerships, deal audits and legislative work. It has partnered with Loti and Vermillio to detect unauthorised uses of clients’ images and voices, issue takedowns, and explore licensing. Chris Jacquemin, WME’s head of digital strategy, told TheWrap, “We’re not a technology company. We’re not trying to be a technology company.”

CAA has taken the other road. It launched CAAvault with Veritone to capture and store authenticated likenesses and voices, and, as The Ankler detailed, put money into Moonvalley so the model itself would be built around consent. The Ankler also noted that Verve has partnered with Incantor AI to track IP rights. Different tactics, same fear: that a face or voice can be summoned without a deal or an agent in the room.

That is why the rights fight sits next to the production story. Hollywood Movietimes has already argued the point in its opinion on AI filmmaking, copyright and consent: the new pipelines only work if someone can say who owns the face on the frame.

A shift, not a funeral

Have US tech and film architects “bypassed” the legacy agencies? Not wholesale. The reporting shows a reroute: studios taking equity in model labs, AI-native outfits building shows for platforms that never needed a talent package, streamers standing up in-house tools, and agencies investing in or partnering with the same technology so they can still clear a likeness and price a star.

The people who understand a model and a story are the ones the industry is now competing for. That is the thesis behind AI Polymath World Cup’s look at the multi-hyphenate advantage inside AI-native studios. The old gate still opens. It is no longer the only door.

Leave a Reply

Your email address will not be published. Required fields are marked *