The studios and the money

The most useful way to read AI filmmaking in 2026 is not as a wave of tools but as a scramble to build companies around them. A distinct layer of AI-native studios has formed, and the pattern is consistent: a credentialed Hollywood or platform operator at the top, a foundation-model partner supplying the pixels, and venture or strategic capital betting that radically lower production costs reset the economics of content. Promise is the clearest example, founded by former Fullscreen chief George Strompolos with an ex-YouTube executive and the AI filmmaker Dave Clark, backed by Andreessen Horowitz and Peter Chernin's North Road. It acquired the AI film school Curious Refuge in early 2025, buying a talent pipeline, added Google as a strategic investor, and stood up a VFXVFX (Visual Effects)Imagery created or altered beyond what was captured on camera, now increasingly generated. division aimed at legacy studios, a sign that near-term revenue may come from selling AI services into traditional productions rather than from its own features.

The other players stake out distinct positions. Bryn Mooser's Asteria partnered with the research firm Moonvalley on Marey, a video model Moonvalley says is trained only on licensed and owned footage, positioned as the legally clean, indemnifiable alternative, a bet backed by an eighty-four-million-dollar round with Comcast and CAA joining. Darren Aronofsky's Primordial Soup is the auteur-branded version, launched with early Veo access from Google DeepMind and reported to be raising about fifteen million dollars. Staircase Studios, founded by a Divergent producer, has targeted roughly thirty projects in four years at budgets of five hundred thousand dollars or less, and Toonstar has run an AI-assisted animation pipeline on YouTube for years, claiming production for fifty to eighty percent less than traditional network animation and a greenlight-to-episode time near ninety days. Those savings are company-reported, not audited, and should be read as directional.

Distribution is where the economics actually resolve, and the honest answer is that most AI film reaches audiences on social platforms and YouTube rather than in cinemas or on major streamers. The exception is vertical microdramaMicrodramaA vertical, phone-first serialized drama told in very short episodes, the format where AI-generated narrative reaches paying audiences first., whose volume-and-localization economics fit generative production almost perfectly, which is why it is the first genuine mass market. The through-line for the book is that capital and distribution, not model quality, now gate this field. The models are good enough that the decisive variables have become who funds the studio, whether the training data and likenesses are cleared, and which channel will actually pay. Expect the near-term money to flow through advertising and microdrama, where cost compression is real and legal friction is low, while the prestige AI feature remains a fundraising story until one clears the copyright, union, and distribution bar end to end.