Cinema Moves Into the Home Office

Four young filmmakers around a director. No cameras, no shooting days, no extras. Instead: screens, a browser window, five thousand image-generation runs — only a fraction of which made it into the finished film. The result was Arena Zero, a ten-minute science-fiction pilot that reached millions of viewers within weeks. Muhannad Nassar from Detroit and Simon Meyer from Berlin, who had never met in person, won 150,000 dollars in prize money with a short film created jointly and entirely on a computer, as part of an international AI film competition with a total prize pool of half a million dollars.

These are the first sober balance sheets of an industry changing its foundations. Film production — for more than a century one of the most capital-intensive, infrastructure-heavy and gatekeeper-driven forms of mass media — is leaving the studio complex behind. In its place, a new economy is taking shape: one in which the story is no longer determined by the size of the budget, but by the strength of the idea.

Why Short Formats Are Made for AI

Anyone speaking of “short-form films” today no longer means only the 15-minute festival short. The term has become a spectrum. Economically, the dominant form is the vertical microdrama, whose unit of length is precisely defined: between sixty and ninety seconds per episode, in 9:16 portrait format for smartphone viewing; a complete season comprises sixty to one hundred episodes.

Alongside it, the cinematic short-film pilot has established itself: five to fifteen minutes, horizontal, narratively self-contained, often serving as a proof of concept for a later series — the very format to which Arena Zero belongs.

Market research firm Omdia estimates global microdrama revenues for 2025 at eleven billion dollars, with a forecast of fourteen billion dollars for the current year. More than 2.3 billion app downloads mark a doubling compared with the previous year. According to Sensor Tower, American mobile users now spend more time on microdrama apps than on Netflix. And on the casting platforms of the American entertainment industry, according to The Guardian, as many as ninety percent of posted roles are already for vertical formats.

It is precisely into this market movement that generative AI enters with a fit that is anything but accidental. A microdrama season requires sixty to one hundred episodes; its profitability becomes visible within a very short time after release. Anyone working with live-action production can barely react to such signals — the material has been shot, the set dismantled, the actors have already moved on to their next contracts.

Those working with AI, by contrast, can rewrite, restage and test a story in different variations before the marketing wave has ebbed away.

The Ukrainian platform Holywater is already pursuing a dual strategy built around this logic. On its platform My Muse, exclusively AI-generated short series are developed as a conceptual laboratory; the most successful stories are then transferred into more expensive live-action production on its sister platform My Drama — where the hit Spark Me Tenderly generated seven billion social-media impressions and twenty million dollars in revenue.

The machine absorbs the risk of the studio decision.

The Systemic Evidence

How consistently this synergy can be scaled commercially is demonstrated by a company from San Francisco: Higgsfield, founded in 2023 by Alex Mashrabov and Yerzat Dulat, reached a valuation of 1.3 billion dollars after its Series A financing round. According to Tracxn, its annualized revenue run rate most recently stood at around 200 million dollars.

In early April 2026, the company launched Original Series, its own streaming service. Its debut was the aforementioned pilot film Arena Zero, directed by Aitore Zholdaskali, followed by twelve trailers. On the platform, viewers vote on which formats should be developed into full series — a form of crowdsourced greenlighting that replaces the studio executive with an audience vote.

Democratization — This Time with Evidence

The promise of film’s democratization is not new. From the 8mm home movie of the 1950s to the DV camera and the mobile phone, it has been invoked at every technological threshold. But now it is becoming measurably more concrete.

The global value chain of content production, estimated by McKinsey at 181 billion dollars, is detaching itself from location advantages such as Hollywood, Pinewood or Babelsberg and beginning to follow talent. Zholdaskali’s four-person team was based in Central Asia when it produced Arena Zero and reached millions of viewers within days. Nassar and Meyer developed their award-winning film across continents, without ever sitting opposite one another — a detail that only a few years ago would have sounded like an anecdotal curiosity, and today increasingly describes the rule rather than the exception.

Stories that would never have received a greenlight under the logic of the classical studio system — because of their risk profile, their niche appeal or their geographic origin — are suddenly finding an audience.

Amid all this acceleration, however, a perception threatens to take hold, one that has already become commonplace in popular reporting: that AI does all of this by itself; that one presses a button and the film simply flows out of the browser. This idea is false — and dangerous precisely because it makes the real value creation of the medium invisible.

A story does not begin with a model, but with a human being who has something to tell. Yet it does not end with that person either. The human being in AI film production carries two roles, which only in popular perception are sacrificed one to the other: they are the initiator of the idea — and the curator of the generated material.

Both roles are indispensable, and the second, contrary to what the imagination might suggest, is often the more time-consuming.

This second role is gaining importance as the industry itself speaks more loudly about a phenomenon that has, in recent months, become known as “AI slop” — the flood of automatically generated, technically polished but dramatically empty moving images increasingly filling platforms such as TikTok, YouTube and Instagram.

YouTube CEO Neal Mohan announced in his annual letter that the platform intended to respond more decisively to the growing flood of low-quality AI content. The very fact that such measures have become necessary reveals the explosive force of the development: what is not filtered through human judgment disappears into the stream of insignificance.

The scarcity economy of attention knows what model providers are reluctant to hear: the value of AI-generated moving image is not a function of its production quality, but of its selection quality.

Arena Zero demonstrates the positive side of this logic in sober numbers. Five thousand generation runs for ten minutes of film — roughly speaking, more than eight attempts for every finished second. Eight attempts, seven of which had to be discarded.

The one who discarded them was not the model. It was Zholdaskali’s trained eye: the eye that distinguishes the right shot from the wrong one, checks character consistency, notices lighting breaks, movement errors and unintended comedy, and senses the rhythm of an edit.

What popular language abbreviates as “AI-generated” is in reality an iterative process of negotiation: the human prompts, the machine generates, the human evaluates, rejects, refines and generates again.

Jonathan Yunger, CEO of the AI film studio Arcana, put it succinctly: “We’re about artist-driven AI, not AI-driven art. If you have a crappy script, it’s just going to be crap.”

He might have added: even a good script becomes worthless if no one can identify the right take among fifty generated variations.

This points to a shift in the most valuable position within the new value chain. It does not lie with the model provider, nor with the platform operator — and not with the machine itself. It lies with the person who knows what they want to tell, which tools they need to tell it, and which of the ten thousand generated images is the one that belongs in the film.

The real bottleneck in AI film production is not the models. It is the trained eyes capable of judging their results without surrendering their judgment.

From Experiment to Professional Production

At this point, it becomes clear that a new professional field is forming around AI-generated moving image. The question is no longer merely which tool can produce the most spectacular images. What matters is who can turn technical possibilities into a narrative, visual and marketable form.

This requires a combination of dramaturgy, visual curation, prompt engineering, art direction, production understanding and legal sensitivity. It is precisely in this in-between space that new specialized providers are emerging, supporting companies, media houses, brands and creative teams as they enter AI-assisted content production.

The AI Media Agency also understands itself as an early Swiss actor in this field. It combines creative concept development with practical AI production — from storyboards, key visuals, avatars and trailers to social-media formats, training sessions and workflow consulting. Not as a replacement for traditional creative work, but as an extension of a craft that is currently being transformed.

Because the easier it becomes to generate images, the more important it becomes to classify them, condense them and place them within a coherent narrative. AI does not make creative judgment obsolete. It makes it more visible.

Risks Remain

Personality rights, training data and authenticity remain unresolved issues that will occupy a generation of lawyers. But structural responses are beginning to take shape: the NO FAKES Act, the EU AI Act, the German collective agreements between BFFS, ver.di and the Producers Alliance, as well as SAG-AFTRA’s agreements with Replica Studios and Ethovox, all mark the gradual construction of a framework designed to keep the technology accessible without dispossessing creators.

For that very reason, AI production cannot be treated as a mere experimental playground. Anyone working with synthetic images, voices, avatars or generated scenes needs not only technical expertise, but also an awareness of rights, transparency, labeling, stylistic coherence and responsibility.

The creative advantage will emerge where technological speed is combined with editorial, aesthetic and legal care.

A Provisional Conclusion

It is the nature of great upheavals that they bring neither the apocalypse feared by their critics nor the utopia promised by their advocates. Sound film displaced silent-film stars, but also brought new ones to fame. Streaming cannibalized cinemas, yet at the same time financed larger budgets for series than ever before.

Generative AI in film will not abolish the actor, but give them new stages; it will not make the studio redundant, but shift its function; it will not disempower the filmmaker, but expand the circle of those who can become one — provided they master both the craft of initiation and the craft of curation.

A fifty-thousand-dollar microseries can reach a larger audience than a forgotten hundred-million-dollar blockbuster. A season of sixty one-minute episodes can carry the emotional arc of a novel without asking more than two hours of the viewer’s time.

As Arena Zero shows, a four-person team can produce a pilot film without a single day of shooting — and replace the greenlighting process of a traditional studio in the time that studio would once have needed for its first internal screening.

The film industry is therefore not facing its abolition, but its most fertile expansion since the invention of the moving image itself.

The vocabulary of this new economy is being shaped right now — by those who understood early that it is not the studio that makes a story, and not the model either.

It is the person who has it. And the person who recognizes it when it appears on their screen.

For Companies and Creative Teams

For companies, publishers, brands, agencies and creative professionals, the question is no longer whether generative AI is changing media production. It already is. The decisive question is how this change can be used competently, meaningfully and with a high level of creative quality.

The AI Media Agency supports organizations on this path — from the strategic classification of generative AI to workshops and workflow consulting, and through to the development and production of AI-assisted film, image and campaign formats.

Anyone who does not merely want to observe this new visual language, but help shape it, needs more than access to the right tools. They need a clear understanding of which story is to be told — and which image truly belongs to it.

#AICreatorDay #AIContentCreation #AIFilmmaking #GenerativeAI #CreativeAI #AIMediaAgency #AIStorytelling #AIWorkflows #CreativeTechnology