Sora Is Not Coming for Your Production Company. Yet.

Last week OpenAI showed Sora, a model that turns a paragraph of text into a minute of video, and my phone filled up with two kinds of messages. Producers asking if we are all unemployed, and tech friends asking why we are not more excited. Both deserve a serious answer, because the demo reel is genuinely the most impressive thing our industry has ever been threatened with.

Let me give you the view from someone who was standing on a dolly track at 6 a.m. the morning the announcement dropped.

What the demos actually show

The clips are remarkable: coherent camera moves, plausible lighting, textures that would cost real money to build or license. Anyone telling you this is just a fancier GIF generator has not watched the woman walking through Tokyo with reflections tracking correctly across wet pavement.

What the demos do not show is control. Filmmaking is not the generation of beautiful moving images. It is the precise repetition of intent across hundreds of shots: this actor, this exact wardrobe, this lens height, matching the shot we made yesterday, cutting against the shot we will make tomorrow. Continuity, direction, and revision are the job. A tool that produces a gorgeous unrepeatable clip is a slot machine, and you cannot shoot a campaign on a slot machine. Nobody outside OpenAI can even touch the thing yet, and the physics still hiccup when you look closely at hands, glass, and anything that collides.

What changes first

None of that means nothing changes. Here is where we expect real impact inside a year or two, and where we are already adjusting:

• Previsualization and pitching. Mood films and animatics that took a week of ripping reference will take an afternoon. This mostly helps directors sell ideas, and it will raise client expectations for pitch polish.
• Stock and B-roll. The generic drone-over-coastline shot is on borrowed time. If your business is undifferentiated footage, diversify now.
• Plates and elements. Backgrounds, skies, and texture layers for VFX work will increasingly be generated rather than sourced.

Notice the pattern: the parts of production that were already commoditized get commoditized harder. The parts that were always about judgment, taste, performance, and trust barely move.

The people question

The uncomfortable truth is that some jobs sit in the commoditized zone, and pretending otherwise helps no one. Our answer at Slipway is to keep pulling our craft up the value chain: stronger directing, stronger creative development, post pipelines that treat generated elements as one more source alongside camera and CG. The crews that treat these models as a department rather than a rival will out-deliver the ones that boycott and the ones that surrender.

And there is a legal fog nobody should ignore. Training data provenance, likeness rights, and commercial usage terms are all unsettled. No sane brand is putting purely generated footage in a national campaign this year, and their lawyers agree with me.

The bet we are making

Audiences are already developing a sense for synthetic. As generated video floods feeds, the market value of the verifiably real goes up: real locations, real faces, real sweat. Ten years from now the premium tier of this industry may be defined by what was actually photographed, the way vinyl and film photography came back as marks of intention.

Sora is a milestone and I refuse to be cynical about it. But the reel that wins the pitch will still be won by an idea, and the model does not have one. That part is still us.