AI video has reached the point where the demos no longer matter. What matters now is economics.

At Techonomy 2025, I moderated two conversations that arrived at the same conclusion from different directions. One focused on synthetic talent and AI-driven media production. The other examined how enterprises are using AI not to replace workers, but to make sense of overwhelming volumes of informationโ€”especially video. Together, they revealed something more consequential than better visuals or faster edits: AI is collapsing production costs, accelerating workflows, and forcing a rethink of who does what in media and creative work.

When the Cost of Video Approaches Zero

Josh Taylor, co-founder and chief production officer of Inception Point AI, came to AI through filmmaking. That matters. Filmmakers understand where the time and money actually goโ€”not into the final shot, but into iteration, revisions, and scale.

Taylor described Inception Pointโ€™s growing roster of AI-generated personalitiesโ€”more than 165 today, with plans to exceed 1,000. โ€œTheyโ€™re characters, yes,โ€ I said during the session, โ€œbut theyโ€™re not people.โ€ Taylor agreed. Instead, he explained the work required to make them legible to audiences.

โ€œWe build character bibles,โ€ he said. โ€œBack stories, flaws, struggles theyโ€™ve had.โ€ The team even laughs about the hardships they invent for these characters, because, as Taylor put it, โ€œit does make it much more relatable.โ€ Want to know what it looks like? Check out Clare Delish, she is smart, cute, and entirely AI. 

That relatability isnโ€™t artistic indulgenceโ€”itโ€™s a distribution strategy. โ€œIt costs us less than a buck to make one of these,โ€ Taylor said. โ€œIf 20 people listen to it, weโ€™re all right.โ€

That single sentence explains why AI video matters economically. When marginal production costs approach zero, the logic of media flips. Content no longer needs mass appeal to be viable. It needs fit the target audienceโ€“even if it is just a handful of people

This is already visible in corporate financials. Klarna told Reuters that generative AI helped cut its marketing costs by $10 million annually, and that 37% of its Q1 2024 sales and marketing savings were directly attributable to AI, while slashing image production timelines from six weeks to seven days.

Mondelez has gone further. The Oreo maker told Reuters it expects generative AI tools to reduce content production costs by 30% to 50%, explicitly targeting creative and advertising workflows.

AI Video Is Not a Tool. Itโ€™s a Stack.

If cost reduction were simply about a single breakthrough model, the story would be easier. But as Jonathan Yunger, CEO of Arcana, explained during the โ€œAI as an Information Allyโ€ panel, professional media doesnโ€™t work that way. โ€œYou need to be working in a multi-model thing,โ€ Yunger said. โ€œRunway is good for one thing, Sora is good for another, VO is good for another.โ€

Yunger was clear about where AI helpsโ€”and where it doesnโ€™t. โ€œWeโ€™re about artist-driven AI, not AI-driven art,โ€ he said. โ€œIf you have a crappy script, itโ€™s just going to be crap.โ€

That distinction is echoed in the data. The OECD defines โ€œexposureโ€ to generative AI as jobs where 20% or more of tasks could be done at least 50% faster using these toolsโ€”and estimates that roughly one-quarter of workers across OECD countries already fall into that category.

Thatโ€™s how the stack works in practice. Script drafting, previsualization, rough cuts, captioning, localization, archive searchโ€”each task gets faster. Together, they move budgets. Phil Petitpont, CEO and co-founder of Moments Lab, focuses on the least glamorousโ€”and most expensiveโ€”part of video production: finding footage. โ€œSeventy percent of the time,โ€ he said, โ€œis just to find the three seconds you need.โ€

Moments Lab uses AI to search massive video libraries and surface relevant clips in minutes. Petitpont described generating a rough cut in under three minutes by turning a journalistโ€™s brief into an agent-driven search across millions of hours of video. Thatโ€™s not about replacing editors, according to Pettipont, itโ€™s about eliminating wasted labor. (And anyone who has slogged through a video library looking for clips will probably agree.)

This is where AI does its quietest, most consequential work: it removes friction from workflows that were never meant to scale.

Productivity Gains, Labor Churn

image

(Fortune Business Insight)

When productivity rises sharply, workforce pressure follows. Not always immediately. Not always cleanly. But inevitably. There seems to be some agreement forming on this. 

  • McKinsey estimates that generative AI could increase productivity in marketing functions by 5% to 15% of total marketing spend, a category that includes video production, editing, and distribution.
  • At the macro level, the IMF estimates that AI will affect about 40% of jobs globally, with advanced economies facing the greatest exposure.
  • The World Economic Forum puts sharper numbers on the churn. Its Future of Jobs Report 2025 projects 92 million jobs displaced and 170 million new roles created by 2030โ€”a net gain, but one that requires massive reskilling and role transition
  • ย Goldman Sachs frames the exposure even more starkly, estimating that generative AI could affect the equivalent of 300 million full-time jobs worldwide through task-level automation

In media and creative work, this doesnโ€™t look like a sudden wave of layoffs. It looks like smaller teams doing more. Fewer junior roles dedicated to logging footage or resizing assets. Higher expectations for those who remain.

The human advantage is accountability, not creativity

Across both Techonomy conversations, one theme kept resurfacing: AI systems donโ€™t own outcomes.Not yet anyway.

Taylor was explicit about the limits of synthetic talent. โ€œWe want them to be helpers, partners, allies,โ€ he said. โ€œNot in a creepy way.โ€ Transparency matters, especially as regulation catches up. New York recently passed a law requiring advertisers to disclose when AI-generated avatars are used in adsโ€”an early signal that the era of frictionless deployment is ending

AI can generate motion, surface clips, and suggest edits. It cannot decide what a brand should stand behindโ€”or what risk itโ€™s willing to take.

Thatโ€™s the real story behind the rise of AI video. It isnโ€™t about machines replacing people. Itโ€™s about machines reordering the economics of creative workโ€”stripping cost and time out of the system, and forcing organizations to confront which human decisions actually matter.

The future of media production wonโ€™t belong to the flashiest model. It will belong to the teams that understand how to build a stack, keep humans in the loop (somehow?), and rides the AI wave without defaulting to AI slop.

Watch the full video here.