Google DeepMind invests $75M in A24 to build AI filmmaking tools

A24 spent years building the rarest asset in entertainment: a logo audiences trust on sight, a promise that whatever is behind it was made by people who cared. So when it accepted a roughly $75 million equity investment from Google and entered a first-of-its-kind research partnership with DeepMind ^[1], the news landed less like a funding round and more like a defection.

The structure is more careful than the headlines suggest. This is an equity stake paired with a multi-year R&D collaboration, not a content-licensing grab, and the agreement explicitly withholds A24's film and television library and data from Google ^[2]. The first concrete artifact is almost anticlimactic: an AI-assisted storyboarding tool, still in prototype, built on DeepMind's Veo video generator and custom models ^[4]. Storyboarding is pre-production scaffolding — the rough visual sketch of a scene before anything is shot — which is exactly the kind of unglamorous workflow step studios are racing to automate. A24's Scott Belsky, who runs its new A24 Labs group, insists the output 'won't look anything like the prompted generation type of AI that people feel uncomfortable with' ^[2].

The irony A24 could not have scripted: its film Backrooms was directed by Kane Parsons, who has said he would erase generative AI 'forever' if he could ^[5]. When one of the studio's own directors is also the movement's loudest skeptic, the deal stops reading as strategy and starts reading as a brand wagering its own identity.

A24 is not an outlier here; it is a latecomer. The studio system has spent two years quietly buying its way into AI production tooling while the public conversation stayed fixated on actors' strikes and copyright suits. Lionsgate struck a custom-model deal with Runway in late 2024 ^[8]; Netflix acquired Ben Affleck's AI filmmaking firm InterPositive in early 2026 to compress production timelines and cost ^[6]; and Amazon MGM stood up 'Project Nara' with AWS and a GenAI Creators' Fund to greenlight AI-assisted series ^[7].

What's genuinely new is the buyer. This is reported as Google's first equity stake in a film studio ^[3]— a search-and-cloud giant taking a direct position in the content business rather than selling tools to it from arm's length. The competitive logic across all of these moves is the same: own the workflow layer — storyboarding, previz, visual effects, distribution — rather than chase the much harder, much more contested goal of generating finished films from a text prompt. Whoever controls the boring middle of the pipeline collects a toll on every production that runs through it, and that is a far more durable position than winning any single 'AI movie' headline.

The strangest term in the deal is the one Google conceded: no access to A24's library or data ^[2]. For a company whose entire AI advantage is built on training corpora, walking away from a decade of distinctive, human-made cinema is a deliberate choice, not an oversight. So what is the $75 million actually buying?

DeepMind's own framing is the tell. Demis Hassabis argues the only way to build tools artists will actually adopt is to 'work directly with them' ^[3], and the partnership is explicitly a multi-year, evolving R&D collaboration rather than a product handoff ^[1]. Read commercially, Google is buying three things data can't provide: a flagship reference customer that makes Veo credible to a skeptical creative industry; a tight feedback loop with working filmmakers to find where the tools break in real productions; and a distribution beachhead in Hollywood without the legal exposure of touching anyone's owned footage. The non-exclusivity on both sides ^[4]reinforces that this is a positioning play, not a lock-in — A24 stays free to work with other AI vendors, and DeepMind stays free to court other studios. Google is paying for proximity and proof, not for pixels.

The backlash arrived within hours, and it landed precisely where A24's value actually lives: its audience. On the film-focused corners of Reddit, the dominant register was betrayal — organized petitions to kill the partnership, loud calls to redirect loyalty to rival distributor NEON, and a recurring, more cynical counter-read that A24 was always a marketing-driven distributor rather than the indie purist its mythology implies. That second strain matters, because it is the only version of the story in which the deal is survivable: if fans decide the 'human craft' brand was partly a marketing construct all along, the contradiction stings less.

The creative community piled on too. Filmmaker Justine Bateman warned publicly that A24 directors should expect their films to be 'altered against [their] wishes' under the arrangement ^[5]. The real exposure here isn't a one-week pile-on that fades by the next premiere. It's that A24's pricing power — its ability to open an unknown film on the strength of its logo — is collateralized entirely by a promise of human authorship. Spend that trust to subsidize Google's R&D roadmap, and the studio risks discovering that the goodwill was the actual asset, and the $75 million was the cheaper half of the trade.