Netflix Paid $587M for an AI Company That Doesn't Generate Video

Last verified: September 1, 2026, directly against Netflix's own newsroom post, its Q2 2026 SEC filing (via TechCrunch), and its Q2 2026 earnings call transcript (via MediaNama).

Netflix paid $587 million in cash for an AI startup, and the tool it bought can't generate a single frame of video on its own. That's the detail most of the coverage buried under the number. InterPositive, the company founded by Ben Affleck that Netflix acquired, doesn't work like Sora, Veo, or Kling. It doesn't turn a text prompt into a scene. It needs your own footage first. Understanding why a company would pay that much for a tool that can't create anything from nothing tells you more about where AI filmmaking money is actually going than the price tag does.

What $587 million actually bought

Per Netflix's own newsroom announcement, the acquisition brought InterPositive's entire team into Netflix, with Affleck joining as Senior Advisor. The deal was announced March 5, 2026, but Netflix didn't disclose what it paid at the time. The figure only surfaced four months later: in its Q2 2026 SEC filing, first reported by TechCrunch on July 19, Netflix revealed it paid $587 million in cash.

What the money actually bought is a tool built around a different idea than most AI video products. Per reporting corroborated by both TechCrunch and No Film School, InterPositive's approach involves ingesting a production's own raw dailies and building a custom AI model trained on that specific footage, one built to speed up VFX rather than generate new footage. Filmmakers then use that model to solve concrete production problems: removing stunt wires, correcting a shot's lighting, reframing a shot that was composed wrong, recovering a shot the crew simply missed, extending or replacing a background. It cannot generate a new scene from a prompt. Affleck himself describes the tool as built with "restraints to protect creative intent," designed for "responsible exploration while keeping creative decisions in the hands of artists." The tool assumes a real shoot already happened.

The bigger picture: one piece of a broader AI push

InterPositive isn't Netflix's only AI investment, and the company has been increasingly open about how much AI already touches its content. On Netflix's Q2 2026 earnings call, held July 16, co-CEO Ted Sarandos said generative AI tools have now been used across roughly 300 titles this year, mostly in post-production, for things like crowd enhancement and historical battle sequences. He specifically called out the docuseries American Experiment, which contains 17 minutes of AI-enhanced footage produced, in his words, "twice as fast and at half the cost of previous options."

Sarandos was also direct that InterPositive is only one part of a larger strategy that includes Netflix's in-house Eyeline unit and its animation lab. And when asked whether AI would start replacing filmmakers, his answer pushed the other direction: "It takes great artists to make something great, and AI is not changing that... Movies are being made by people who make movies. AI provides them with better tools to make them even better." He also said the cost savings from AI-assisted production would likely get reinvested into making more content, not cut from budgets outright.

What this actually signals for AI filmmakers

A recent post on this blog covered Sora's shutdown and the shift toward generate-from-a-prompt tools like Veo and Kling. InterPositive sits in a different category entirely, and that's exactly what makes this deal worth paying attention to if you're building your own AI filmmaking workflow rather than just watching the industry from outside it.

The generate-from-nothing tools get most of the hype because the demos are dramatic: type a sentence, get a shot. But the $587 million question is which category is actually where studios are putting real money for real production use, and the answer, at least at Netflix, is the tool that needs your footage to already exist. That's not a small distinction. It means the current center of gravity in professional AI filmmaking isn't "skip the shoot," it's "shoot, then use AI to close the gaps a real production always has": the missed shot, the wrong light, the wire that shouldn't be visible.

For anyone directing AI-assisted work at a smaller scale, that's a useful signal to actually act on:

Real capture still matters, maybe more than ever. A dailies-trained model like InterPositive's is only as useful as the footage it's trained on. If the underlying skill studios are paying for is "AI that extends and fixes real production," then knowing how to shoot usable raw material, even rough or minimal, stays valuable in a way that a pure prompt-to-video workflow doesn't require.

The most fundable AI tools right now solve specific, boring production problems. Not "make an entire film," but "fix the coffee cup that moved between takes" or "get the shot we didn't have time to grab before the light changed." Building AI-assisted skills around narrow, concrete post-production fixes is a more Netflix-shaped bet than building toward full generative replacement.

Watch where studios put their own money, not just what gets the most demo views. Sora dominated headlines and then shut down. InterPositive got comparatively little attention next to the flashier generate-from-prompt launches, and it's the one a major studio paid $587 million for. The tools getting real budget aren't always the ones getting the most social traction.

FAQ

How much did Netflix pay for InterPositive, and when? Netflix paid $587 million in cash. The acquisition was announced March 5, 2026, but the price wasn't disclosed until Netflix's Q2 2026 SEC filing, reported by TechCrunch on July 19, 2026.

What does InterPositive's AI tool actually do? Per reporting corroborated by TechCrunch and No Film School, it ingests a production's own raw dailies to build a custom AI model, then uses that model to fix production problems in post: removing stunt wires, correcting lighting, reframing shots, recovering missed shots, and extending backgrounds.

Can InterPositive generate a video from just a text prompt? No. Per Netflix and Affleck's own description of the tool, it requires a production's actual footage as its training base and cannot create a scene from nothing, unlike prompt-to-video tools such as Sora, Veo, or Kling.

Is Netflix's AI use limited to InterPositive? No. On its Q2 2026 earnings call, Netflix said generative AI has been used across roughly 300 titles this year, and co-CEO Ted Sarandos said InterPositive is one part of a broader AI strategy that also includes Netflix's in-house Eyeline unit and its animation lab.

---

This is the same lesson we apply to every project: AI tools are most valuable when they're solving a specific problem in a real production, not replacing the production itself. If you want an editing partner who thinks about AI the same way, book a call, or read about what happened when Sora shut down for the other half of this picture.