Netflix paid about $587 million in cash for Ben Affleck’s AI startup. It completed the acquisition back in March 2026, but disclosed the price only in its regulatory filing for the second quarter. Earlier reports had estimated the value of the deal at up to $600 million. With this sum, Netflix is buying the layer between filming and the screen.
The model learns only from its own filming
The startup is called InterPositive, and its tools rest on a model trained exclusively on the dailies of a specific film or series. According to TechCrunch, it fills in missing shots, corrects lighting, changes or improves backgrounds and adjusts the framing of existing footage. Work that until now meant reshoots or hours of manual post-production is handled directly on material the studio already has.
Training only on the studio’s own dailies carries both technical and legal value. The studio trains the model on footage it holds the rights to, so the question of where the training data came from, which courts are wrestling with for large generative models, does not arise. For deployment in Europe, where regulation is pushing for transparency about training content, a model closed over a single production is a clean answer.
Post-production moves in-house
The whole InterPositive team is moving to Netflix, and Affleck stays with the company as a senior adviser. On its second-quarter earnings call, Netflix said it had used generative AI on around 300 titles this year. The acquisition adds both its own tool and the people who built it to that deployment.
Fixing footage therefore moves at Netflix from an external supplier to in-house infrastructure. Whoever owns the layer that repairs the footage decides the cost of each production in the catalogue. Post-production, which studios used to buy as a service, becomes an asset at the biggest streamer.
Comments