A24 and Google's AI Partnership: The Money Isn't the Part Worth Noticing
By Dan Gardner | June 23, 2026
Google has invested $75 million in A24 to develop AI tools for filmmakers.
A24 hardly needs an introduction — Moonlight, Everything Everywhere All at Once, and Hereditary all came out of this company. For the past decade or so it has been the least studio-like studio in America: small budgets, director-first, strange subject matter, living off one piece of reputation — that its films are not assembly-line product.
Which is why this money landing on A24 has caused far more noise in America than the number itself would suggest.
And the genuinely interesting part of the story probably isn't the money. It's how American audiences reacted.
First, the Mood Over There
This section may be the least familiar part of the whole article for a Chinese reader: in America, the general public attitude toward AI is scepticism, and in places outright hostility.
A survey this year from the Pew Research Center (an American public-opinion research organisation) found that among American adults under 30, roughly half believe AI will do harm to society. Note the word — not "carries risks," but "harm."
The American internet has also grown a term of abuse that is now completely standard: AI slop. Rendered literally into Chinese it comes out as "AI 泔水" — AI pig-swill. It refers to things churned out in bulk with AI: cheap, samey, obviously made without anyone caring — images, articles, video, music, all of it. Today the phrase is an insult, and a fairly heavy one. To call something AI slop is to say it doesn't deserve to be looked at.
The crucial thing is that these two facts collided: A24's brand is built precisely on the people who find that stuff most repellent. Its core audience is under 35, picky about films, and identifies itself through its taste — exactly the group that trusts AI least in Pew's numbers.
So when the news broke, a lot of people's first reaction was a sigh: even A24 is going to start making AI slop now?
That's the working environment for American filmmakers today. Any move that touches AI has to clear the audience first — not the regulator, the audience. A Chinese reader doesn't have to share the sentiment, but knowing it exists is what makes each of A24's next steps legible.
What A24 Says for Itself
A24's Scott Belsky said something worth noting, because it sounds as though it was aimed at exactly that group.
His point: the AI companies got the pitch wrong from the very start. What they sold was "cheaper and faster" — the two words creators least want to hear, because translated they mean "you can be cut." The genuinely valuable use would be to keep creative control and support risk-taking, and what gets built should look nothing like the prompt-in, content-out AI that makes everybody uncomfortable.
In other words: what they want to build isn't a generator. It's a tool.
The Best AI Is the Part the Audience Never Sees
This is my own view, and the only direction I think this investment is worth betting on.
The work AI should be doing in film all happens before the cameras roll, and not one frame of it ends up in the finished picture.
- Storyboards — draw a scene out before shooting it, try a few ways of covering it
- Casting boards — see in advance what a given set of actors feels like in the same scene
- Art direction and sets — what has to be physically built, and what will never be in frame
- Script analysis — whether a character stays consistent, whether motivation has gaps, whether the arc holds
- Location scouting — what the same place looks like in a different season, at a different hour
- Stunt previs — work out the safety and the cost first, before anyone actually jumps off it
- Script breakdown — mark up every prop, location, and day-or-night scene, and build the call sheet
- Footage management — finding one specific shot inside three hundred hours of material during the edit
All of these jobs have one thing in common: every one of them is thinking-it-through work, not making-it work.
Not a single one replaces an actor, a writer, or a cinematographer. What they replace is waiting — waiting for a sketch, waiting for a spreadsheet, waiting for someone to watch three hundred hours of footage from the top. And in the film business, waiting is money, and money is "we can't afford to try that idea."
That line is a clean one. What audiences are worried about is AI displacing creation, and nothing on that list is creation.
I Should Declare My Position
RunPee is made by me alone. One person writing the code, doing the design, answering the user email. I've done it myself for twenty years.
Sometimes this app's income doesn't even cover the grocery bill. I can't afford a translator, can't afford a designer, can't afford a second engineer.
Since AI arrived, I get more done in a month than I used to in a year.
That isn't ad copy, it's my actual situation. So my view on this is biased, and I'll own the bias: I'm the kind of person these tools pulled back from the edge. What a big studio saves with AI is budget. What I get back with AI is whether the thing can keep going at all.
Where Caution Is Warranted, It Really Is
I'm not going to overstate the case.
AI will bring two things at the same time: some people will survive because of it, and some will lose their jobs because of it. The medical benefits and the occupational damage aren't an either/or — they'll happen together, and they usually land on different people. The people who are worried aren't failing to understand. They're running a different set of numbers than I am.
One more thing: whoever says today that "AI only does pre-production" may well find in two years that the line has been pushed somewhere else. Technical boundaries don't stop because somebody made a promise.
So the thing actually worth watching isn't how much money is in this investment, but what A24 builds with it: a tool that gives creators a few more chances to fail, or one more content machine.
A24 is betting on the former. Given its audience and its reputation, it can't really afford to bet on the latter anyway.
Related Reading
- A Sister and a Brother — RunPee is made by two people
- What Is a Peetime — what a Peetime is