Emmy-winning creator Bernie Su on why YouTube filmmakers succeed in Hollywood — and what AI will do to the movie industry next

YouTube, Hollywood, and the AI wildcard

Bernie Su has won three Emmys — for Lizzie Bennet Diaries and Emma Approved, the first two primetime Emmy wins for a YouTube show, and Artificial, the first Emmy win for a Twitch show, which also won a Peabody. He now runs Pickford, where he's building AI-powered interactive narrative experiences for theater audiences. His read on the YouTube-to-Hollywood pipeline, and where AI fits into it, cuts against most of the breathless coverage.

The trend isn't new

The wave of YouTube creators breaking into Hollywood box office — Obsession, Backrooms, Iron Lung this summer — looks like a turning point from the outside. Su says it's a ramp that's been building for at least five years, visible to anyone paying attention at events like VidCon. MrBeast, he points out, is now a Prime Video creator, not a YouTuber in the traditional sense. The platform label has already become secondary to the franchise.

What changed isn't the direction of travel, it's the execution. Earlier attempts to plug a creator with 10 million followers into a studio film failed because they were casting an audience, not a brand. Markiplier's Iron Lung works because it's authentically his — the same creative identity his audience already follows, just on a bigger screen.

“You had this era where there were these YouTuber-led movies — the movies probably weren't regarded as great movies, but what missed the mark there was that they just kind of plugged and played a dude in there. Now with Iron Lung and Markiplier, it's his. It's very much him being him... Hollywood is still the best place to amplify your franchise. We're building at Pickford — a detective series where the audience could interact with it in real time because the AI was writing the show as it goes. Literally impossible without the technology.”

The franchise play

Hollywood's structural advantage isn't prestige, it's amplification. Su frames it as franchising: a breakout film becomes the launch pad for toys, merch, sequels, video games, and spin-off series, and Hollywood still has the machinery to scale that faster than any individual creator team. K-pop Demon Hunters is his clearest example — it became a franchise overnight, and now Hollywood is scrambling to catch up with the ancillary demand it created.

YouTubers, though, are more nimble. Smaller teams, faster decisions, less bureaucracy. Su's expectation is that the healthiest creative arrangements going forward involve studios showing creators how the franchise machine works, then largely stepping aside.

Horror as the low-budget wedge

Horror keeps producing these breakout moments — Blair Witch, Paranormal Activity, now Obsession — for compounding reasons. Fear is culturally universal in a way comedy isn't. Productions don't need recognizable stars (which actually helps: unknown actors make audiences feel the threat is real). And the format doesn't require the location costs or CGI budgets that romantic comedies or sci-fi demand. Su expects Obsession knock-offs to follow, just as found-footage films flooded the market after Paranormal Activity. Whether any of them are good is a separate question.

The AI movie question

Asked when a film with 90% of its budget in AI compute crosses $30 million at the box office, Su doesn't have a clean answer — and he's honest about why. The Amazon-backed Hicksville (he notes he hasn't seen it) is the closest current candidate, but his instinct is that it won't get there. Two things have to align: the models have to produce something genuinely entertaining, and the public ritual of shaming AI-made content has to cool off enough that audiences don't reject it on principle. He can't predict the timing on the second one.

His cleaner argument is that the most interesting AI films won't be AI-cost-reduced versions of existing genres. They'll do things that were structurally impossible before. Pickford's current project, demoed at Sony last week, is a detective series where the AI writes the show in real time as a theater audience interacts with it on their phones. The experience is launching at Alamo Drafthouse locations around the country this year. It won't hit $30 million, but Su's point is that it couldn't exist without the technology — which is a different and more durable creative case than shaving VFX costs.

Where the blue sky is

Su is most animated about two ideas: personalization and consequentiality. Personalization means geolocation-triggered customization — a horror film that knows you're in Pasadena, or generates a welcome sign for your town on the fly. Consequentiality means the audience feels it changed something, not just watched something. He ties this to the parasocial relationships Gen Z and Gen Alpha already have with their favorite creators: they show up to Markiplier's film partly because they feel a relationship with him. The next step is making them feel like they influenced the story itself.

AI is well-suited to both, he argues, because speed and real-time optimization are exactly what personalization and interactive narrative require — and those are genuine AI strengths rather than areas where the sloppiness of current models is most exposed.