The $29.95 Commercial: Inside the strange new economics of AI video and the creative labour nobody is counting.
The sentence arrives casually now, almost as a reflex.
It shows up in pitch meetings, brand reviews, procurement calls, and late-stage budget negotiations. Someone leans back in their chair, glances at the production estimate, and says it with a kind of satisfied clarity:
“If we shot this properly, it would cost two hundred thousand. But with AI… it’s basically free.”
Sometimes there’s a laugh. Sometimes a pause while the implication settles in. Sometimes,One of my scene construction interfaces for a client project.One of my scene construction interfaces for a client project. a quiet sense of technological inevitability — the feeling that an old cost structure has finally been defeated. Two hundred thousand dollars reduced to a software subscription.
A commercial priced like a streaming service.
No one says it explicitly, but everyone understands the meaning: the economics of production have collapsed.
The camera has been replaced by the prompt. The set by the LMM model. The crew by the interface. And if that is true — if the physical machinery of filmmaking has dissolved into computation — then an uncomfortable question follows naturally.
What exactly are we paying for now?
The number that refuses to die
The figure itself — the mythical $200,000 commercial — has become a kind of rhetorical anchor in marketing culture. It represents the world before AI. The world of cranes and lighting rigs and location permits and crew call sheets printed before sunrise.
It is shorthand for friction.
For decades, producing a piece of commercial film meant organising reality at scale. People had to travel. Equipment had to be transported. The weather had to be anticipated. Schedules had to be synchronised. Insurance had to be purchased in case anything went wrong, which it frequently did.
Money moved because the physical world resists coordination.
A production budget, in other words, was not just a creative expense. It was a logistical negotiation with physics.
AI has removed much of that negotiation. But something else has taken its place.
What disappears when the trucks go away
The most visible change in AI production is the absence of things that once defined filmmaking: no lighting trucks idling outside a location, no cables taped across concrete floors, no assistant directors shouting over walkie-talkies while a clock quietly converts time into money.
Scenes emerge instead from iterative generation. Landscapes materialise without scouting. Camera movement is simulated rather than executed. The sun behaves exactly as instructed. From a distance, this looks like elimination — the disappearance of cost itself.
But elimination is the wrong word. What actually happens is migration.
The responsibilities once distributed across dozens of specialists do not vanish. They condense. They relocate. They accumulate inside a much smaller space, sometimes inside a single operator.
A cinematographer once shaped light through lenses and physical placement. Now someone must define, describe, calibrate, and stabilise that light across hundreds of generative outputs.
A location manager once secured environmental continuity. Now someone must ensure that a fictional environment remains coherent across model interpretations that do not inherently remember what came before.
An editor once assembled footage captured in time. Now someone must determine which among hundreds of possible visual realities becomes canonical.
The work does not shrink.
It internalises.
The quiet expansion of effort
Inside AI production environments, something strange happens to time.
Physical filmmaking is governed by scarcity. Daylight fades. Crew overtime accumulates. Constraints force decisions forward.
Generative production removes most of those pressures. Variation becomes cheap. Alternatives multiply instantly. Possibility expands faster than consensus can form.
A shot can be regenerated endlessly — not because the previous version failed, but because a marginally better version might exist.
And so teams keep looking.
They compare emotional tone, micro-expressions, atmospheric density, camera drift, colour temperature shifts so subtle they would never survive discussion in a traditional edit suite. The work becomes an extended act of aesthetic discrimination — searching for an image that feels inevitable rather than merely acceptable.
Iteration becomes the dominant activity. Not shooting. Not rendering.
Choosing.
And choosing, when options are infinite, is exhausting work.
The Explosion of Iteration
This is the part almost nobody talks about. When every shot is cheap, decision pressure multiplies. You don’t shoot 5 takes. You generate 200 variations.
Traditional production is constrained by physics. You cannot shoot endlessly. You cannot move lights forever. You cannot reschedule the sun.
Scarcity forces decisions. AI removes those constraints completely.
When every shot costs almost nothing to generate, iteration becomes infinite. An infinite number of possibilities creates a new cost centre: decision fatigue at an industrial scale.
Instead of choosing between five takes, you are reviewing two hundred. Instead of approving a lighting setup, you are selecting between aesthetic micro-variations that differ in emotional tone by fractions of a degree. Instead of committing to a direction early, teams postpone commitment because they can.
Cheap generation creates expensive hesitation.
Creative convergence — the process of narrowing toward a final expression — becomes the dominant labour in AI production. And convergence is cognitively demanding work.
The new labour nobody is pricing correctly
Agency rate cards were designed for a different world. They itemise hours of production, days of shooting, equipment rentals, post-production workflows. They are built to price tangible coordination.
Traditional production budgets have always been dominated by material logistics — the cost of organising people, equipment, locations, time, and physical execution. AI production changes that balance entirely. The centre of gravity shifts away from the orchestration of matter and toward the orchestration of thought. What begins to dominate instead is cognitive systems work: conceptual architecture, visual world logic, prompt design at narrative scale, model selection and testing, style continuity, multi-tool coordination, generation supervision, version control, review governance, narrative integrity, and brand risk management. The physical machinery recedes. The mental machinery expands.
But what happens when the tangible coordination dissolves, leaving only sustained cognitive supervision?
Someone must hold the narrative intact while images fluctuate. Someone must decide when variation stops and authorship begins. Someone must protect brand meaning when visual possibility becomes effectively limitless.
These are not technical services. They are acts of judgment. Judgment is slow. Judgment is fragile. Judgment cannot be parallelised. Yet it is precisely this labour — the continuous shaping and stabilising of meaning — that becomes central in AI production.
And it rarely appears on an invoice.
Why the illusion persists
Software pricing creates a powerful psychological distortion. Tools announce their cost clearly. Human cognition does not.
A generative platform charges a monthly fee. The number is visible, measurable, and defensible in a budget meeting. The expanded effort required to guide that platform toward coherence is invisible — absorbed into creative time that organisations have never learned to price properly. So the story becomes simple.
The software is cheap. Therefore, the production is cheap. Therefore, the creative work must also be cheap.
It is an elegant narrative. It is also wrong.
The risk beneath the optimism
When organisations believe production has become trivial, they behave accordingly. They compress timelines. They reduce oversight. They assume that variation guarantees quality.
But abundance does not produce coherence. It produces noise.
Brand meaning — already difficult to maintain in fragmented media environments — becomes even more vulnerable when image generation outpaces narrative control.
Visual consistency drifts. Emotional tone fluctuates. Symbolic intent blurs. Nothing fails catastrophically. Everything simply becomes slightly less precise.
Which, over time, is how craft erodes, and slop takes center stage.
The economic inversion
The central shift in AI production is not technological. It is economic. Filmmaking was expensive because manipulating matter was expensive. Cameras, lights, locations, and physical presence — these were the dominant cost drivers.
Now the manipulation of matter is trivial. The manipulation of meaning is not.
The industry has not yet adjusted to that inversion.
It continues to talk about the disappearance of equipment while ignoring the emergence of interpretive labour — the continuous work of deciding what an image should express, what a sequence should imply, what a brand should feel like when the constraints of physical reality no longer guide those decisions.
What the $29.95 commercial really costs
The subscription is real. The computational efficiency is real. The logistical simplification is real.
But the work of authorship — the sustained act of shaping possibility into intention — remains exactly as demanding as it has always been. In some environments, more so.
The camera did not carry that burden before.
People did. They still do.
The cost of producing images has collapsed. The cost of deciding which images deserve to exist has not.
Until the industry learns to price that distinction honestly, the $29.95 commercial will continue to circulate — not as an economic reality, but as a comforting misunderstanding about where creative work actually lives.
And misunderstandings, unlike production budgets, have no upper limit.
Sooner or later, every brand and every agency will confront the same realisation: the hard part is no longer making the video. The hard part is deciding what the video should be — and knowing when to stop.
That requires new disciplines. New workflows. New decision structures. New forms of creative supervision that most organisations have never had to build before, because physical production constraints once did that work for them.
Those constraints are gone now.
Which means AI production isn’t a cost-saving tool. It’s an operational system that has to be designed, governed, and learned. Teams need to understand how to manage generative iteration, how to maintain aesthetic continuity across models, and how to protect brand meaning under conditions of infinite variation.
None of that comes bundled with the subscription.
It has to be built.
Some companies will figure that out through painful trial and error. Others will decide that building those capabilities deliberately — with guidance from people who have already navigated the shift — is faster, cheaper, and far less risky.
Either way, the era of treating AI video as a cheap production shortcut is ending. The era of treating it as a creative systems discipline is just beginning.
If your organisation is navigating this transition — building AI production capability, designing workflows, or training teams to work at generative scale, then we should talk. That is the work I do at RockPaperScissors.