Updated October 24, 2026. No newer funding filing or announcement from Primordial Soup turned up in a search on this date, so the July raise below is still the latest money news. What this update adds is the studio's running output, a weekly series on TIME's YouTube channel, in a new section before the close. This is an analysis of public reporting, not a review of anything I worked on.
What is Darren Aronofsky's AI film studio?
On July 22, 2026, TheWrap and Variety reported that Primordial Soup Labs filed with the Securities and Exchange Commission for a $15 million equity raise. The filing lists Darren Aronofsky and his producing partner Dylan Golden as executive officers, and it says the company had already sold just over $11 million of that total, with the first sale dated July 1. So the AI film studio a lot of people treated as a curiosity last year is now a funded company with real money behind it and a paper trail in a federal filing.
The name on the door is the part worth sitting with. Aronofsky directed Requiem for a Dream, The Fountain, Black Swan, and The Whale. His entire body of work is about obsession, the body, faith, the parts of being human that do not optimize. He is, by any reasonable definition, an auteur. And he is building an AI film studio.
I do not read this as an observer. I have directed and edited commercial work since I founded Pichorra Filmes in São Paulo in 2012, and I use AI video tools on real jobs now, Kling and Seedance and Gemini Omni, with clients and deadlines attached. So when an auteur raises fifteen million to build inside the same tools I ship with every week, I read the filing the way I read a call sheet. As a set of decisions someone already made.
The short film almost nobody read correctly
Primordial Soup did not start with a fundraising round. It started with a movie. The studio launched in May 2025 through a partnership with Google DeepMind, and its debut project premiered at the Tribeca Festival on June 13, 2025. The film is called Ancestra. It runs about eight minutes. It was directed by Eliza McNitt, and it combines live-action footage with sequences generated by Veo, Google's video model.
Here is the detail that the "AI slop" reflex skips. Ancestra is about a child born with a hole in her heart, and McNitt built it from the real circumstances of her own birth. It is not a tech demo dressed as a story. It is a personal film, the most personal thing a director can make, with the AI doing the part a human hand could never shoot: the cosmic, the interior, the impossible camera inside a body.
That is not a machine replacing a filmmaker. That is a filmmaker using a machine to reach an image she could see and could not otherwise capture. The authorship did not leave the room. It decided where the tool went.
Why an auteur is the least surprising person to build this
The common story says AI comes for the artist first, and the serious filmmakers hold the line. Aronofsky just raised eleven million dollars arguing the reverse, and I think he is right, for a reason that has nothing to do with technology.
Cinema has always been auteurs adopting machines early. Kubrick chased lenses NASA built so he could shoot Barry Lyndon by candlelight. Cameron pushed digital compositing because the image in his head did not exist yet on any shelf. The directors with the strongest personal vision are usually the first to reach for a new instrument, because they already know exactly what they want the instrument to do. A vision that specific is starving for range.
The filmmaker with nothing to say has no reason to learn a new tool. The one with too much to say and never enough budget has every reason. I learned that on a stop-motion campaign I directed for Forno de Minas, through WMcCann, where every frame was a decision made before the camera rolled and revision was almost impossible. What I wanted then was more range for the vision, not a way to skip the vision. That is the same hunger a Veo sequence feeds when it lets a small team reach an image that used to require a studio.
The reception is split, and both sides are half right
The coverage of the raise divided fast. Gizmodo ran the money news under a headline about a company that "makes very bad AI movies." The trades reported it straight, as a serious filmmaker attracting serious capital. Both reactions are pointing at something true.
The skeptics are right that a lot of AI film output is genuinely bad. The quality gap between what these tools promise and what most people ship with them is real, and I see it every day. But bad output is not evidence that the direction is wrong. It is evidence that most of the output has no author behind it. The problem was never that a neural network rendered the frame. It was that nobody decided anything before it did.
Which is exactly why the name on the filing matters. This is not a faceless content farm. It is a director with a body of work going back to the nineties choosing to build here, using a hybrid model where the human decision stays load-bearing and the AI handles the impossible shot. That is the version of this future worth taking seriously, and it is the version most of the mockery is not aiming at.
What has Primordial Soup made so far?
Two pieces of work are public, and they are the best evidence for or against the thesis of this article.
- Ancestra (2025). The eight minute short directed by Eliza McNitt, premiered at the Tribeca Festival on June 13, 2025. Live action combined with sequences generated by Veo.
- On This Day... 1776 (2026). An animated short form series about the American Revolution. TIME Studios distributes it on TIME's YouTube channel, one episode a week through 2026, each timed to the 250th anniversary of the event it shows.
The credits on the series are the interesting part. According to TIME's announcement, Aronofsky is executive producer, a writers room led by Lucas Sussman writes it, artists animate it with generative tools made in part with Google DeepMind, SAG actors voice it, Jordan Dykstra composes the score, and a post team edits, mixes and grades it. Variety led its coverage with exactly that detail: an AI studio that hired human actors for the founding fathers.
Read that list as a call sheet and the AI occupies one line of it. Every other line is a department with a person in charge. That is the hybrid this article argues for, and it is also an answer to the people asking whether an AI studio means nobody gets hired. On the one production we can inspect, the writers, the voices, the music and the edit stayed human.
Whether the episodes are good is a separate question, and the reception has been rough, including the Gizmodo headline quoted above. You can judge for yourself, because the series is free on YouTube. Watch one episode with the credits in mind. Ask which decisions in the frame came from a person and which came from a default. That exercise teaches more about directing these tools than any benchmark.
What the $15 million buys
Money does not buy taste. That is the part the headline number cannot capture. Fifteen million dollars buys compute, staff, and runway. It does not buy the thing that made Ancestra land, which is a director who knew that a hole in a newborn's heart was the story before a single frame was generated.
This is the same lesson that keeps repeating across every AI story I write about here, from Netflix calling generative AI infrastructure to Google putting $75 million into A24. The tool keeps getting stronger. The tool never gets wiser. It amplifies whatever an author hands it. Hand it a real decision and it can amplify something that looks like ten times your budget. Hand it nothing and it amplifies the nothing, in 4K, at scale.
Aronofsky is not betting that AI can make a film. He is betting that a filmmaker, armed with AI, can make more of the films only that filmmaker could make. Not because the technology stopped needing a human. Because the human was always the part that decided what the technology was for.
I take the same bet on a smaller scale on my own timeline, which is why the vocabulary I bring to these tools matters more than the tools themselves. The fifteen million is a headline. The auteur showing up is the whole story. Everyone worried that AI would erase the author just watched one of the most uncompromising authors alive raise the money to prove the opposite.
Sources: TheWrap (Jul 22, 2026); Variety; Deadline; Variety (2025, Primordial Soup + DeepMind); Tribeca Festival; TIME (Jan 29, 2026); Variety (2026, voice actors); Gizmodo
Sources and verification
Published Jul 24, 2026. Facts verified against TheWrap, Variety, Deadline, and The Hollywood Reporter (Jul 21-22, 2026) for the SEC filing; Variety, Tribeca Festival, and Google DeepMind's own blog (2025) for the Ancestra short and the DeepMind partnership.
Corrected Jul 25, 2026: the $11 million was raised, not spent, and the raise opened July 1, 2026, so the "spent a year" framing was wrong. The SEC document is a Form D notice of an exempt offering, not a term sheet. Aronofsky's feature work begins with Pi in 1998, so "thirty-year body of work" was overstated.
Updated Oct 24, 2026: added the On This Day... 1776 series, checked against TIME's distribution announcement and Variety. "Fourteen years" replaced by the verifiable founding year of Pichorra Filmes.