DeepMind puts $75M into AI tools for Hollywood movies

Google DeepMind & A24 (which produced movies such as ‘Everything Everywhere All at Once’ & ‘The Backrooms’) have entered into a long term research collaboration around using AI in filmmaking. As part of this collaboration, Google is investing around $75 million into A24. This is an interesting twist on traditional AI deals, where instead of deploying a completed piece of AI technology into a process, it creates a framework whereby practicing artists can trial AI tech during active production cycles and feed back insights to the researchers.

1Real films as a research environment

This is framed as a research collaboration, rather than a commercial deployment. As Eli Collins, VP of Product at Google DeepMind puts it, “this gives us valuable feedback from real world experts.” Filmmakers at A24 will use AI tech as part of their normal production cycle, not in a lab, but on real productions with real constraints. They won’t be sharing details of the exact tools they are using or what they hope to achieve, but by investing $75 million, Google is getting access to feedback from people who will use the tools in a way that matters.

A24’s involvement here isn’t just about branding. A24 is renowned for taking risks on weird and wonderful projects, and their body of work is widely seen as artistically challenging. They represent a different type of user to a studio specializing in franchise films. If AI tech is found to be helpful in an A24 production, it will have to convince other filmmakers that it improves results in qualitative ways, not just quantitative ones (ie. making things faster or cheaper).

2What this signals for AI in creative work

This aligns with a trend we’ve been seeing from large AI labs looking for structured feedback from practitioners in fields where the end result is subjective and processes are hard to replicate in a lab environment. We’ve seen this in medicine, law and architecture. The aim is always the same — give practitioners access to the tech so researchers can understand where it goes wrong and where it goes right, not just how it performs in benchmarks.

If you’re building creative AI tools, this suggests that future generations of those tools will be informed by feedback from the most demanding users. When DeepMind releases research based on this collaboration, it will tell us what happens when AI meets a filmmaker under pressure to make decisions, not a controlled experiment. We’ll learn more about where AI can be helpful, and where it interferes or creates outputs that feel off somehow.

Finally, Google’s willingness to invest $75 million is a sign of their ambitions. This isn’t a sponsorship agreement or joint marketing campaign — it’s a financial investment in a studio in return for access to a testbed. Google wants to understand what AI can do in filmmaking, at the high end of the market where it has to compete with humans.