Cinema is not dying because audiences stopped caring about film. It is dying because the people running it stopped making decisions that could produce anything worth caring about. That is a business problem, not a cultural one - and it has a clear and traceable cause.
How You Optimize a Creative Industry to Death
Every major studio is ultimately answerable to capital. That is not unusual - most industries are. What is unusual about film is that capital and creative quality are in unusually direct tension. A film that takes a genuine risk might return ten times its budget or nothing. A sequel to an established property will almost certainly return two or three times its budget, reliably, with predictable variance. For a public company managing quarterly earnings, the choice is not difficult. You make the sequel.
The problem is that this logic, applied consistently across an entire industry over twenty years, does not produce a stable business. It produces a slow erosion of the thing that made the business worth having in the first place. Audiences do not go to cinemas to see sequels to sequels. They go because occasionally something extraordinary happens on a screen and they want to be in the room when it does. Strip out the extraordinary, and you have stripped out the reason to show up.
The Franchise as a Business Model
The Matrix is a precise case study because the original trilogy was, by design, a complete thing. Three films with a beginning, a middle, and an ending. The story resolved. The characters finished where they were going. Whatever criticisms one might make of Reloaded or Revolutions, the trilogy had the structural integrity of something that knew it was going to end. That is increasingly rare, and it matters - because an ending is what separates a story from a content pipeline.
The correct response to owning the Matrix IP after the trilogy was either to leave it alone or to build something genuinely new inside the same universe. The world the Wachowskis constructed is large enough. New characters, a different era, a different corner of the machine war, a different question about the nature of the simulation - any of these could have been the basis for something that respected the original while standing on its own. A current audience does not need Neo. It needs a new conflict with real stakes and a story it has not already seen.
What Warner Bros. produced instead was The Matrix Resurrections in 2021 - eighteen years after the trilogy ended - with the same lead actors, the same Agent Smith, and a plot that literally resurrects the characters the previous films had concluded. The decision to bring back the same faces was not a creative one. It was a calculation that name recognition and nostalgia would lower the perceived risk of the investment. The studio did not believe in the universe. It believed in the brand. The audience noticed. The film failed - not because the Matrix concept is exhausted, but because the film offered nothing except proof that the people making it were not willing to find out whether it could work without a safety net.
Star Wars industrialised this dynamic at a scale that was previously unimaginable. The acquisition of Lucasfilm was not a creative decision - it was a content library acquisition. What followed was the systematic conversion of a mythology that had genuine cultural weight into a production pipeline. Films, series, spin-offs, prequels, sequels, all released on a cadence driven by release windows and subscriber targets rather than by whether anyone had something worth saying. The audience eventually signalled, clearly, that it was exhausted. The response was not to slow down - it was to try different combinations of the same components.
A franchise is not a creative decision.
It is a risk management strategy dressed as one.
Marvel and the Pipeline Fallacy
The Marvel Cinematic Universe is the most important case study in modern film economics because it succeeded spectacularly for long enough that the people running it drew the wrong conclusion from their own success. From Iron Man in 2008 through to Endgame in 2019, the MCU produced something genuinely unusual: a connected series of films that were, individually, mostly good, and collectively something audiences had never experienced before. The shared universe worked. Characters crossed over. Threads built across years. When it paid off, it paid off at a scale that reshaped the entire industry's assumptions about what a film could be.
The mistake was in identifying why it worked. The shared universe was not the product. It was the reward for the product. Audiences invested in the connections between films because the films themselves had earned that investment. Iron Man was a good film. The Winter Soldier was a good film. Guardians of the Galaxy was a good film. Each one worked as a standalone experience, which meant that when they intersected, the intersection meant something. The connective tissue had weight because the individual pieces had weight.
What Marvel - and Disney, which had acquired it - took from this was a different lesson: that the pipeline itself was the asset. That releasing four films and multiple Disney+ series per year, all tagged as MCU content, would sustain audience engagement indefinitely because the brand had accumulated enough goodwill to carry anything attached to it. This is a category error. Goodwill is not a renewable resource. It is the residue of past quality, and it depletes every time it is spent on something that does not deserve it.
The post-Endgame MCU inherited an audience that had spent eleven years being rewarded for paying attention. What it offered in return was volume. Films that were competently assembled but existed primarily to move pieces into position for the next film. Disney+ series that were mandatory viewing if you wanted to understand the cinema releases, effectively converting a leisure choice into homework. Characters introduced and abandoned. Storylines launched and quietly dropped. The shared universe, which had once been the payoff for engagement, became the reason engagement was required just to follow along. The audience began to disengage - not because superhero films stopped being possible, but because the implicit contract had been broken. The pipeline had forgotten that it was dependent on the films being good, not merely connected.
Streaming did not fix this dynamic. It accelerated it and added a new and more corrosive incentive on top: subscriber acquisition. A studio making theatrical films needs a film to be good enough that people will pay to see it. A streaming platform needs content to exist in sufficient volume and variety that enough different people will start a subscription and not cancel it. These are genuinely different problems, and the second one does not require quality - it requires surface area.
Altered Carbon is a personal example of what this looks like from the other side. The first season was genuinely good - ambitious production design, a coherent adaptation of its source material, the kind of science fiction that takes its own premise seriously. I liked it. I kept watching. I was exactly the subscriber Netflix should want to retain. The show was cancelled after two seasons regardless, and I eventually cancelled the subscription. Those two facts are more connected than they might appear.
The issue is not that Netflix made a bad decision by its own logic. The issue is what its logic measures. A subscriber who genuinely loves three or four ambitious series and watches them carefully registers identically to a subscriber who half-watches twenty shows and never finishes any of them - as long as both keep paying. The metric that matters is new sign-ups, because that is what the market rewards. Retention of engaged viewers who care about specific content is a secondary concern at best. Cancelling Altered Carbon did not cost Netflix the number it was optimizing for. It cost them me - eventually - and I was never the unit being counted.
What this produces is a library optimized not for quality or loyalty but for the minimum threshold of good enough to keep someone paying while the next acquisition drives new sign-ups. The service does not need you to love it. It needs you to not quite hate it enough to cancel. That is a meaningful distinction. A library built around that incentive looks very different from one built around making things worth watching - and over time, the difference becomes visible.
| Model | Primary incentive | What it produces |
|---|---|---|
| Theatrical studio | Maximise opening weekend and franchise extension potential | IP acquisitions, sequels, safe casting, risk-averse greenlight decisions |
| Streaming platform | Subscriber acquisition and retention across the broadest possible audience | High volume, uneven quality, early cancellations, content churn |
| Franchise extension | Extract value from an established property at minimal creative cost | Diminishing quality per instalment, audience fatigue, brand erosion |
| AI-assisted production | Reduce per-unit content cost to near zero while maintaining surface coverage | Structural collapse of the floor — unlimited slop at no marginal cost |
AI Is Where This Was Always Going
The question AI puts to the industry is not a technical one. It is a choice: do we hire a writer, or don't we? Do we commission a director with a specific vision, or do we generate the output that vision would have produced? That choice is now explicit in a way it was not before. And given everything the industry has already demonstrated about its priorities - the franchise logic, the subscriber metrics, the systematic elimination of creative risk — there is not much reason to expect most studios and platforms to choose the human when the alternative is cheaper and more controllable.
This is what artists have been worried about, and they are right - but not quite for the reason usually given. The concern is often framed as AI producing bad films. The more accurate concern is that AI produces films optimised for the same thing the industry already optimises for: volume, familiarity, minimum acceptable quality. AI does not introduce a new set of values into the pipeline. It enforces the existing ones more efficiently. A model trained on what has performed before will reliably produce more of what has performed before. For an industry that has spent two decades treating proven formula as the safest bet, that is not a warning. It is a feature.
What gets lost is not quality in the narrow sense - an AI can produce something watchable, and watchable is already the threshold most of the industry is aiming for. What gets lost is the possibility of something genuinely unexpected. A writer or director brings a specific perspective that did not exist before, that cannot be interpolated from what came before, and that occasionally produces something the market did not know it wanted until it arrived. The first Matrix was that. The early MCU was that. The pipeline, fully enforced by AI, eliminates the conditions under which that becomes possible. Every film becomes a variation on the centroid of everything that already worked. The ceiling does not disappear - it just gets permanent.
What Would Fix It
The honest answer is that the incentive structure would need to change, and there is no obvious mechanism by which that happens from inside the industry. The studios that consistently produce the best work are either small enough to be insulated from franchise logic, backed by individuals with enough capital and conviction to override short-term return calculations, or operating in markets where the economics of Hollywood do not apply in the same way.
Audiences have some leverage. Theatrical attendance is a direct signal. Not watching a streaming series past its first episode is a direct signal. The platforms and studios read these signals carefully - they just tend to read them as evidence that they need better marketing rather than better films. That misreading is itself a symptom of an industry that has fully internalised the idea that the product is secondary to the pipeline.
Cinema at its best is one of the few experiences that cannot be fully replicated on a phone screen with half your attention elsewhere. The industry has spent twenty years making content that can be. That is not an accident. It is what you get when every structural incentive points toward volume, safety, and extraction - and nobody in a position to change it has a strong enough reason to try.