Hollywood and GenAI: Slouching Towards Acceptance
Sustaining Innovation is Not Enough to Stave Off Disruptive Change
Artificial intelligence won’t kill Hollywood. But it will transform the process of making movies and TV shows. There’s a difference between destruction and disruption. Let’s take a closer look.
In this article, I want to peer beyond the exaggerated claims of the total destruction of the traditional media industry. Instead, let’s consider how lassitude and inertia are likely to leave the biggest players in an established industry vulnerable to a rising wave of change that anyone can foresee.
Today, major media companies are making minimal efforts to prepare for the changes heralded by generative AI, which means, once again, they will be caught flatfooted, just as they were caught out by the advent of streaming video, YouTube, social media, TikTok and micro-dramas, which together have chiseled away a cumulative 50% of viewing time from traditional media in less than 20 years.
This is not a story about ignorance. The executives who manage the motion picture studios and TV networks are fully aware that generative AI has the potential to reshape the industry. Some of them are already using it. But the overly cautious and tentative way major media companies are using AI guarantees they will be blindsided by competitors who use it differently.
Today’s article refers to Clayton Christensen’s framework from The Innovator’s Dilemma to explain why this is likely to occur in the media industry. And it uses the example of the Honda/Harley-Davidson collision of the 1970s to help us anticipate what will come next.
The TL;DR version of my argument:
Hollywood is moving through the classic stages of grief about the demise of a 100-year-old production process: denial, anger, bargaining, depression. Some signs of acceptance are beginning to appear, but the process is uneven and slow.
Christensen’s framework describes the strategic choice that every studio now faces: sustaining innovation or disruptive innovation. That is, should we use AI to preserve the old production process? Or use AI to invent an entirely new process? Most media companies will opt for the former. That’s a strategic blunder.
What the major studios are doing with generative AI today consists of the minimal amount of sustaining innovation. Going through the motions. They are bolting AI onto the periphery while leaving the core production model untouched. This is a tradeoff between near-term comfort and long-term pain. Retooling the entire motion picture pipeline with AI will be difficult. The longer they wait, the harder and more urgent the transition will become.
Fecklessness is a byproduct of faulty incentives. Executive decision-makers inside major media companies operate within an incentive structure that favors an abundance of caution. Those who advocate for a low-risk approach face zero career risk. However, those who recommend adopting new technology bear the risk of significant penalties for anything less than stellar implementation. This is why most executives recommend a slow and cautious approach, rather than early adoption.
Meanwhile, AI-native competitors, unburdened by such caution, are moving at full speed to devise a new production process leveraging the rapid improvement in video models; and they are connecting with a new audience. Today, AI workflows for filmmaking remain somewhat cumbersome and unreliable, but they are improving rapidly. By the time the major media companies respond, some AI-native firms will likely be established with fresh story franchises, a devoted fan base, and a hyper-lean production process. And once again, big media will be forced to play catchup.
The window to respond is closing faster than anyone in Hollywood is willing to admit.
This situation is a microcosm of a major transition that is playing out across the entire economy. Large established enterprises in every sector are struggling to reorganize their legacy business process around AI, while small, nimble AI-native startups intend to scale up fast by leveraging the unique affordances of the new technology. This is true in many industries, not just media. It’s a footrace between big companies retooling around AI and startup companies scaling with AI to steal market share.
1. Hollywood’s Dramatic Stages of Grief
In 1960, a Harley-Davidson owner would have laughed at a Honda 50cc Cub. It was considered a toy motorcycle. Harley fans used a derisive term to dismiss Honda bikes: “rice burner”. You couldn’t take it on the highway. The little bikes didn’t sound right, didn’t look right, and the people who rode them weren’t motorcycle kind of people.
That attitude did not age well. Honda crushed Harley in two decades.
By 1980, Harley-Davidson’s share of the American heavyweight motorcycle market had collapsed from 75% to 25%. The company nearly went bankrupt. In 1981, thirteen of its own employees decided to buy the ailing firm back from AMF, the conglomerate that had acquired it, for $80 million, just to keep the venerable motorcycle manufacturer on life support.
By then, Honda had already won a commanding share of the motorcycle business.
Honda didn’t beat Harley by building a better Harley. It built a different motorcycle for a different kind of rider: someone who had never ridden at all. It started with the 50cc Cub in 1959, then moved those riders up to 125cc, then 250cc, then to the 750cc in a relentless progression up the power pyramid.
By the time the CB750 arrived in 1969, Honda had millions of loyal riders who felt no attachment whatsoever to Harley-Davidson. They hadn’t grown up dreaming about Harleys. Honda was the brand they identified with. I’ll come back to this.
Let me pause here. Before I finish that story, I want to describe what’s happening in Hollywood right now, because the emotional dynamic is familiar. I’ll join these two threads at the end.
Faced with the long-term secular decline of its traditional revenue sources, and confronted by the advent of generative AI, Hollywood’s motion picture industry is moving fitfully through a grieving process. They are mourning the demise of the old, tried-and-true, 20th century model of filmmaking as it fades into history.
As my friend Mark Pesce reminds me, Elizabeth Kübler-Ross identified five stages of grief: denial, anger, bargaining, depression, and acceptance. Thanks to AI, according to Pesce, who is only half kidding, now most software developers and some professionals in Hollywood are obliged to move through a grieving process to mourn their loss of an identity derived from skills that have been rendered obsolete.
Like any mourner, the folks in the motion picture industry are not moving through these stages in a tidy sequence. It’s ugly and upsetting. All of the early stages are visible simultaneously, right now, in the trade press and in board rooms and in the layoffs and in the watering holes of Hollywood.
Denial
Denial is mostly behind us. Three years ago, the prevailing belief was that AI could never produce a compelling script or movie. In 2023, the Writers Guild strike committee declared that AI was a plagiarism machine and incapable of useful contribution to screenwriting. That position has since collapsed. Two months ago, the WGA closed a new agreement with the AMPTP that accepts AI as a normal writer’s tool, with the proviso that writers — not producers — control its use.
Anger
Anger is loud and current. Hannah Einbinder, star of HBO’s Hacks, delivered her verdict on AI filmmaking at a press conference in West Hollywood last month:
“The people who make this stuff are losers. They’re not artists. They’re not creative. You guys suck. No one likes you. You are a loser. You will never be cool. And you probably had a rolly backpack in high school. I want to put your head in the toilet and flush.”
Way to keep it classy, Hannah.
The show’s creators, Paul W. Downs and Lucia Aniello, back up Einbinder fully. They told WIRED they personally won’t work with anyone who uses AI in any creative capacity. The Hacks production has banned AI from the show entirely.
Jen Statsky, who created the series with Downs and Aniella, expressed moral outrage about AI, “I wish that I believed it was in better hands, but I don’t. And until there’s guardrails put on, until there is some stoppage mechanism to make sure that we are protecting the humanity both in art and people’s livelihoods, it ain’t good.”
These folks are not outliers. Many prominent actors, including Scarlett Johanssen, Tom Hanks, Stephen Fry, Jenna Ortega, Cate Blanchett, Joseph Gordon-Levitt, Sean Penn, and Whoopi Goldberg have decried the use of AI, and their views are echoed by many of their 170,000 peers in SAG-AFTRA and by other creative professionals.
This anger is understandable. It comes from craft pride and a reasonable suspicion that studios will use AI to cut salaries. Plus, the conviction that rogue deepfake creeps will clone the voices, works, and likenesses of talent and famous celebrities.
These industry-specific concerns also play into the general suspicion that AI is a kind of heist; it will intensify income and wealth inequality that benefits an already-rich tech elite. Downs said it plainly to WIRED: “The thing that it’s benefiting, especially in film production, is the top 1 percent... the shareholders who get to say, ‘We don’t need a VFX house.’”
Anger as a strategy has limits. It grabs headlines. It telegraphs an anti-technology stance that may bolster credibility among creative peers. But it does nothing to stop the technology.
The most that anger will accomplish is to bully into silence those who are considering how to use AI. Thereby, perhaps anger will hobble the fledgling technology initiatives of the movie studios and TV networks, on whose survival those actors, showrunners, and their managers depend to earn a living.
No media executive wants to incur the wrath of celebrity talent. Few are willing to put a target on their backs by openly championing the use of this controversial technology (I’ll point out a few brave pioneers later in this article).
Moreover, the actors’ anger, aside from Einbinder’s venom, tends to focus on very specific issues; but that precision vanishes in the blurry haze of hatred towards all things AI. Legitimate professional concerns about non-consensual deep fakes expressed by Ortega, and the non-permissioned cloning of voice or likeness expressed by Johanssen, Blanchett and Gordon-Levitt, get mingled into the broad but unspecific resistance towards AI that has emerged as a populist rallying cry across the US.
Consequently, the nuanced position of the actors is lost in the noise. Their message has been eclipsed by crude cartoonish anti-AI polemics. Studios do not seek to incite the ire of the masses, especially by favoring a still-somewhat-unreliable technology that currently does more for Big Tech more than it does Big Media.
Bargaining
Meanwhile, behind closed doors and away from the press, quiet progress is happening. Bargaining is occurring in contract language and academy rules.
SAG-AFTRA’s new tentative 2026 TV/Theatrical agreement with the AMPTP permits fully synthetic AI-generated actors in signatory productions. This is not a blanket license to displace human actors. The agreement contains certain conditions: producers must prove that a synthetic performer delivers “significant additional value” that a human actor couldn’t provide. Notice and negotiation are required. But the door is now open to synthetic actors in a way it was not last year. As Eriq Gardner wrote in Puck News after reading the 18-page agreement: “Hollywood has effectively agreed to that future.”
Nine months earlier, SAG-AFTRA had issued a statement dismissing the AI persona Tilly Norwood as “not an actor.” But by May 2026, the union had conceded the terms under which Tilly Norwood’s and its equivalents can work on signatory productions. The negotiation is no longer about whether synthetic actors will be included in film productions. It’s about price. That means it’s certain to happen, despite the hue and cry from some prominent actors.
The Academy of Motion Picture Arts and Sciences recently updated its Oscar eligibility rules to ban AI-generated performances and AI-written screenplays from award consideration. But AI tools remain eligible in VFX, sound, and editing categories, provided they operate as assistive tools. The most prestigious institution in American cinema has drawn a line at authorship while accepting the technology everywhere else. That’s not a ban. It’s a negotiated settlement.
Notice the contrast between what gets press attention and what gets done at the bargaining table. Einbinder’s outburst grabbed headlines in every tabloid. But the SAG-AFTRA concession was buried in the trades. The anger is quotable, but the contracts are real. Follow the money.
Finally, Acceptance
Acceptance is beginning to appear, unevenly, sometimes in unexpected places.
Steven Soderbergh, a filmmaker who often experiments with new technology, partnered with Meta’s AI team to generate 10% of the shots in his recent documentary John Lennon: The Last Interview. The generated shots did not include a deepfake of Lennon. Instead, AI was limited to animated imagery to accompany abstract commentary in the Beatle’s voiceover.
Soderbergh was aware that his use of AI will make him a lighting rod for critics, admitting “I am going to have to wear this.” (Maybe he should watch out for a sneak attack by Hannah Einbinder next time he uses a public restroom at a premiere.)
By crossing the Rubicon, Soderbergh has successfully shifted the discussion away from the binary “dare I use AI or not?” towards the topics of taste and artistic discretion that include AI as one option in a filmmaker’s palette of tools.
Soderbergh enjoyed the experience so much, he is doubling down on the technology, indicating that he plans to use “a lot of AI” in his next project about the Spanish American war.
Paul Schrader wrote Taxi Driver, Raging Bull, and The Last Temptation of Christ. He is not a tech founder, nor is he the stereotypical early adopter. He spent his career at the highest levels of American screenwriting. In January 2025, Schrader publicly praised ChatGPT’s ability to generate original film ideas in seconds. He argued that AI script coverage is superior to human coverage because it lacks the sycophancy bias of studio readers. By October 2025, he was predicting the first fully AI-generated feature film was just two years away. He said he has the perfect script for it.
This week, Shrader will deliver the keynote presentation at AI On The Lot, the largest conference on generative AI in film and media, now in its fourth year, drawing more than 1,200 attendees, including executive from every major studio.
To summarize: denial is waning, but the anger still burns hot. Bargaining is happening quietly. And the acceptance is already starting to arrive piecemeal, but it’s not yet evenly distributed, to paraphrase William Gibson.
Let’s consider how this is likely to play out in the near future.
2. Three Paths
While the bigwigs of Tinseltown are distracted by the emotional upheaval of Hollywood’s Kübler-Ross journey, they may overlook a more consequential strategic challenge: the path to adopting generative AI in filmmaking presents a textbook example of Clayton Christensen’s theory of disruptive innovation.
The Innovator’s Dilemma (1997) is about what happens to dominant companies when new technology arrives. Christensen observed a consistent pattern. Incumbents don’t ignore new technology. They adopt it. But they adopt it in a specific way: they use the new tech to improve their existing product for their existing customers.
Christensen called this sustaining innovation. That’s a fancy way of saying incumbents deploy just enough technology to preserve the past, but not enough to make a drastic change for the future.
The problem is that sustaining innovation leaves a gap wide open for a new rival that is unconcerned about preserving the old business model. New technology often enables a different product, at a much lower price, aimed at a customer segment that the incumbent failed to serve. A new entrant will build that product. It may be inferior to the incumbent’s product by every traditional measure, so the incumbent doesn’t worry about it. It’s beneath their notice.
Then, gradually, in Christensen’s account, the new entrant’s product improves. It climbs the market. By the time it reaches the incumbent’s core customers, the new entrant has already built a massive base of loyal users who feel no attachment to the old brand.
Christensen called this process disruptive innovation.
Today the term “disruption” is so overused that we tend to forget Christensen’s original meaning. In 2016, at an event where we were both guest speakers, Christensen admitted to me that “the theory has become a victim of its own success” because lazy pundits have co-opted the word disruption to describe any sort of technological change.
What Clay meant by the term is specific: a low-cost, de-featured product that appeals to a new customer segment that was previously ignored by incumbents and their expensive, feature-laden products. Disruption consists of blowing open the bottom of the market to a new customer base.
In the 1960s, disruptive innovation gave Honda a beachhead into Harley’s dominance in the motorcycle market. In the 2020s, it will do the same for a fresh crop of AI-native motion picture producers who intend to steal a chunk of audience attention from big budget films and TV series.
Three paths are available to Hollywood. The studios can do nothing and ignore AI; or they can use AI to preserve the existing production model, by trimming costs at the margins while leaving the core process intact. Or they can use AI to build an entirely new production model that is 100x to 1000x more efficient.
I’ve described the defects in the old model in previous articles. Here’s the short version: the 20th-century assembly-line process invented by Thomas Ince is a left-to-right linear sequence in which creative decisions are locked at each stage. The people with the most visibility into the finished product, including the editors and VFX supervisors in post-production, are the least empowered to change it, even if they know how to save a flawed film. Since 1997, the average crew on a top-200 feature film has grown by more than 70%, and most of that new hiring is concentrated in the disempowered post-production phase. In the same period, box office ticket sales are down 47%. When labor cost rises and revenue falls, it is long past time for a new method.
AI promises to invert the old assembly line process of film production. With generative AI, scenes can be iterated endlessly, swiftly. Characters can be replaced late in the process. Settings and plot points can be altered after the story is locked. The cone of creative possibility expands as production proceeds, rather than narrowing as it does by necessity in the assembly line approach to filmmaking. It is a categorically different way to make a film. It won’t be easy for big studios and networks to adopt this new approach, but this is what the second century of filmmaking looks like.
The path that preserves the assembly line is sustaining innovation. The path that replaces it completely is deeply disruptive.
3. What the Studios Are Actually Doing with AI
The motion picture companies must choose among three scenarios.
The first scenario is to do nothing: let the legal team set the terms, avoid any AI application that could trigger a headline or a guild action, and wait.
The second scenario is to experiment at the margins: AI for script coverage, scheduling, de-aging, cheap VFX, editing, marketing. Useful, cost-saving in small ways, and safe because this sort of AI won’t ruffle any celebrity feathers.
The third scenario is to commit AI to the core of the production process itself, completely retooling filmmaking for a non-linear iterative workflow, knowing that this transition will be uncomfortable and the first results may be imperfect.
Most studios are doing a bit of the first and some of the second. That’s why AI adoption in Hollywood today mostly consists of sustaining innovation by Christensen’s definition.
AI for script coverage. AI for scheduling and budgeting. AI for de-aging actors. AI for VFX shots. AI for editing trailers. These are legitimate uses, and they are also insignificant. None of them will change the fundamental production process. None of them will address the core problem of cost, which stems from the growing size of the crew, the inexorably rising cost of principal photography, and the rigid left-to-right industrial process that Tom Ince built.
The examples are instructive. Metaphysic used AI to de-age Tom Hanks and Robin Wright in Here, directed by Robert Zemeckis. The production cost still clocked in at about $50 million, filmed conventionally on sets with a full crew. The AI was a tool applied to solve a specific problem within an unchanged process.
Amazon’s House of David grew from roughly 70 AI-generated shots in early episodes to more than 350 across the second season, using Midjourney, Runway, Magnific, and Topaz. Director Jon Erwin argues this approach can save LA film jobs by making large-scale visual effects affordable on television budgets. He’s right that it reduces cost somewhat. But what Erwin is describing is textbook Scenario Two: AI bolted onto the conventional production process in order to preserve it, not the comprehensive re-imaging of that process to replace it.
The Russo brothers hired a machine-learning scientist from Apple to guide AI adoption at AGBO, their production company. Callaia, an AI tool noted in the New York Times, reads scripts and generates 35-page coverage reports including historical comparisons and suggested release patterns. Minor tweaks to the existing process.
Sony Pictures has stated it will use AI to cut production costs and improve efficiency. At the same time, Sony Music has forced the removal of 135,000 AI-generated deepfake tracks from streaming platforms and has opted its entire catalog out of AI training. Same parent company, two different responses to the same technology. Use it where it’s efficient. Fight it where it threatens existing revenue. That is Scenario Two with a legal defensive posture.
These are real developments, and they all illustrate the point that AI is currently being applied to isolated tasks rather than to the larger project of retooling the entire workflow of producing and shooting a film or TV show.
The pattern across the industry is consistent: deploy AI only where it avoids conflict with guilds, agents, talent relationships, and legal exposure. Which means deploying it everywhere except the places where it would make the most difference.
Where AI is most likely to be deployed under this logic is in post-production, namely video effects and computer-generated imagery. That’s the one department where studios can cut cost without triggering a labor strike or damaging a relationship with a powerful director or star. The problem is that VFX shops have been absorbing budget cuts for two decades. Intensifying that pressure will drive more of them out of business, reducing production capacity across the industry.
The worst outcome of Scenario Two is the false comfort it creates. Studio heads can point to their limited AI deployments and say: “Look, we are adopting AI in an industry-friendly way.” Except what they’re adopting is the minimum amount of AI in a way that changes nothing important. No significant costs will be reduced; the schedule won’t be shorter. Film crews and budgets will continue to be bloated.
Metaphors of rearranging deck chairs come to mind.
4. Why the Incentive Structure Guarantees This Outcome
This is not a story about stupid people. The executives who lead TV and film companies are shrewd. They respond rationally to incentives.
To understand why Big Media companies are slow-rolling AI adoption, we must consider how the decision to use these tools is made.
Who, exactly, wields the veto power over the use of generative AI?
The answer is complicated because veto power is diffused through layers of advisors and executives who are more concerned with risk mitigation than embracing the future. None of these people are responsible for budgets or workflow.
The in-house legal counsel warns vaguely about copyright exposure, liability, and the murky prospect of obtaining a copyright on the finished product: they do not offer clarity or legal solutions, instead they generate a fog of uncertainty that leaves the rest of the organization in limbo about how to proceed. The best they can offer is a yellow light. Then the public relations team piles on more doubt, expressing worries about the potential for negative press coverage. Then talent relations managers speculate about the fraying of relationships with the agents and managers of celebrity talent. Labor coordinators raise the prospect of breaching guild contracts. Story development teams with no firsthand experience repeat horror stories about AI-generated narratives.
Any executive who feels inspired to adopt AI must contend with this kind of institutional whisper campaign. Anyone who wants to use AI will find themselves shadow boxing against vague, unproven fears and dark speculation.
None of these people enjoy any upside benefit from the adoption of AI in filmmaking, but each of them faces a clear downside risk. If AI adoption leads to a lawsuit, a bad headline, a talent boycott, or a guild action, then the responsible executive will take the blame.
On the other hand, if the studio fails to adopt AI today and thereby is on the wrong side of disruption a few years hence, the chorus of doubtful executives who urged a cautious approach will be long gone. They leave no fingerprints on the murder weapon.
The problem is an asymmetrical alignment of incentives. It’s no surprise that there are few vocal champions of AI in most media companies.
What I’ve learned in three decades of launching new digital services is that every new initiative depends on a champion: someone to lead the charge, who owns the responsibility for ignoring the naysayers and doomers, and who tackles the job of launching the new thing, risk be damned.
I’ve launched new media businesses, including TV networks, streaming services, and game platforms, across Asia, Europe, and the Americas. Each time, as the launch deadline approached, some well-credentialed expert would arrive with data and charts explaining why the project was doomed to fail. My team was instructed to ignore the experts and proceed with the launch. Every one of those launches succeeded. The “experts” just moved on to advise some other client.
Experts pay no price when their forecast is wrong. Their job is to identify risk. They are not paid to see opportunity. They won’t get a bigger bonus if a producer seizes the opportunity and makes a hit show.
The studio heads of production are surrounded by merchants of doubt. Senior executives who deal with legal affairs, PR, talent, and labor. People far from the actual production who wield a soft veto in a whisper campaign. All of them are adept at spotting risk. None of them is accountable for the cost of inaction.
The cost of inaction is real. When an industry is on the brink of being redefined by a radical shift in economics, doing nothing is not the safe option. It is a very slow version of the riskiest option of all: staying mired in the past.
Which is why, when I speak to the heads of production at motion picture studios, they all seem to tell me the same thing. They punt the decision to the directors: “We don’t tell our filmmakers what tools to use. We won’t tell them to use AI. Just as we don’t tell them what kind of camera to use.”
On the surface, this answer seems like a respectful acknowledgement of artistic integrity and remarkable self restraint. But it’s really just a weak dodge. No director or producer is going to retool the studio’s production process without a clear mandate.
Consider it from the filmmaker’s perspective: after a year of development hell, your project has just been greenlighted by a studio. Are you now going to risk it all on a new kind of technology like generative AI? No way. Your budget includes enough money to hire a cast and crew and shoot on a proper sound stage. Why would you do anything other than produce the film in the tried-and-true way?
The director has no incentive to save money or experiment with a novel production method, unless she is driven by the same creative curiosity that attracted Steve Soderburgh. You can count those directors on both hands.
The problem is that nobody in the entire hierarchy of a motion picture company truly owns the responsibility for radical innovation. The inevitable result is tiny, incremental tweaks on the periphery of the production process.
5. Here Come The Disruptors
While Hollywood dithers by tinkering with AI at the margins, other people are building a new film production process with AI at core. That’s what makes this a footrace.
In China, production companies are making AI-generated micro-dramas for $30 per finished minute. No actors, no crew, no camera, no sound stage. In March 2026 alone, nearly 50,000 AI-generated micro-dramas were uploaded to Douyin, China’s version of TikTok. That single month’s total nearly matched the platform’s entire output for all of 2025. The Chinese market for AI micro-dramas is projected to exceed $3 billion this year.
In previous articles, I’ve stated the cost comparison in detail. Here’s a snapshot. US studio television drama costs roughly $5 million to $10 million per finished hour, which works out to $83,000 to $166,000 per minute. A low-budget independent feature might run $5,000 to $20,000 per minute. The Chinese AI micro-drama producers are at $30 a minute. (That’s not a typo).
That’s not a 100x cost advantage. At the studio end, it’s closer to 3,000x.
The content is currently aimed at Chinese audiences watching on cheap smartphones. The stories are simple. The production values are modest. Harley-Davidson owners would have recognized this product immediately. They probably would have slagged it off as another “rice burner.”
But here’s what the research tells us about the audience being built. By 2020, more than 2.2 billion people globally were watching short-form video regularly, with Generation Z accounting for 54% of those users. In Asia-Pacific, where the market is concentrated, 89% of TikTok content is user-generated. This audience has never expected professional production values. They have been raised on something other than Hollywood.
The Chinese AI micro-drama producers are not competing for Marvel’s audience nor HBO’s. They are building a new audience pyramid from the bottom, aimed at viewers that the big US studios have never tried to serve. Some of those viewers are in China. Many others are in Southeast Asia, India, Africa, and Latin America. Some are in the United States.
The path to the US market is already mapped. Language localization — re-voicing content in American English with AI-modified voices — will be the first step, achievable at tiny marginal cost when the content was AI-generated to begin with. Replacing synthetic Chinese-looking actors with synthetic American-looking ones is the next step. I’ve described this mechanism in detail in a previous article as hyper-localization. The thin end of the wedge is language. The fat end of the wedge is full content substitution, market by market.
On the more visible end of the disruptive spectrum, the video AI model company Higgsfield screened Hell Grind at the Cannes Film Festival this month: at 90 minutes, a complete long-form narrative with zero live-action footage, the first fully AI-generated feature to screen at a major film festival.
Meanwhile, director Chuck Russell, whose films include The Mask and Eraser, with combined global box office exceeding $1 billion, has announced a partnership with Higgsfield for two AI-driven sci-fi features deploying AI tools across the entire pipeline.
The disruptive side is not one thing. It’s a many-headed hydra. It spans fully synthetic productions and high-budget hybrid films. It spans Chinese smartphone content and Cannes festival entries. It also includes promotional films, industrials, advertisements and movies sponsored by marketers. Every AI-generated film will compete for eyeballs and will split a fragment of audience attention away from big-budget studio fare.
What these AI productions have in common is a novel production method: these firms are evolving away from the labor-intensive, left-to-right assembly-line process that Hollywood has relied on for a century. In this, they enjoy a head start that could turn into a durable long-term cost and speed advantage.
6. The Honda Pyramid
Back to the opening story about Harley-Davidson versus Honda. In the 1950s, Harley-Davidson motorcycles were sold to leather-clad punks and tough guys whose hero was Marlon Brandon in The Wild One.
In 1959, Honda entered the US market.
Honda didn’t just make motorcycles. Honda built a new kind of rider. The 50cc Cub in 1959 was aimed at exactly the people who would never have walked into a Harley dealership: housewives, teenagers, college students, and commuters who had never considered motorcycles at all.
Sales languished in the low tens of thousands until 1963 when a UCLA college student proposed a new tagline: “You meet the nicest people on a Honda.” The spot aired during the Rose Bowl. In 1963, Honda sales spiked from 40,000 units sold to 150,000. By 1965, Honda sold 1 million and captured half the US market.
Honda then moved those riders up the product ladder to bigger and fancier bikes. By the time the CB750 arrived in 1969 — 736cc, electric start, hydraulic disc brakes, 67 horsepower, but priced more than $1,000 below comparable Harleys — Honda had millions of loyal riders who had no cultural attachment to Harley-Davidson whatsoever. Harley’s heavyweight market share fell from 75% in 1973 to 25% by 1980.
Harley’s legal response to Honda warrants a couple of paragraphs. In 1995, when Honda introduced the Shadow ACE, a cruiser with a single-pin crank specifically engineered to produce a loping V-twin sound, Harley sued. Not for copying a design. There was no patent on the V-twin or the single-pin crank. Harley sued for the sound, claiming it as a trademark. The lawsuit lingered in the courts for years. Honda kept building and selling the Shadow ACE throughout. Eventually Harley dropped the case. Honda quietly retired that model and moved on. Neither side won anything material.
Now look at the media industry. The major studios, publishers, and celebrity talent have sued AI companies for copyright infringement. The studios have also negotiated contractual restrictions through guild agreements, and lobbied for legislative protection. Some of those battles will produce real outcomes. But as a strategy, it is the same move Harley made. Litigation buys time. It does not build a better motorcycle. While the lawyers fight, the competitor keeps shipping.
Meanwhile, the audience being built by Chinese AI micro-drama producers does not read the Hollywood trades. Their preferences are being formed right now, outside the studio system, on platforms the US media giants do not control. Some of these new viewers won’t particularly care for HBO’s dark and meanspirited programming, including snarky shows about self-absorbed comedy writers who can’t find work.
By the time Hollywood notices this new audience at scale, it may be too late to reach them. By then, the micro-drama producers will be working their way up to TV shows and feature-length films, if there is still an audience for longform.
This is the motion picture industry’s version of the Honda pyramid. It is being built right now, at $30 a minute, 50,000 uploads a month, aimed at an audience that has never had a strong attachment to Hollywood product.
7. Don’t Count on Big Media Winning the Next Round Intact.
Hollywood will survive this. I’ve argued that point in detail elsewhere. The studio system is a network of networks: financing relationships, distribution infrastructure, accumulated craft knowledge, and institutional capacity for managing creative risk at scale. Those networks are resilient. AI will not dissolve them, but it might reconfigure them.
But survival is not the same as continuity.
Harley-Davidson survived Honda’s advent on American shores. Harley is still in business. It holds about 40% of the US heavyweight motorcycle segment today, but now it operates in a market flooded with excellent Italian, German, and Japanese competitors. That survival required a near-death experience, a complete management change, and a decade of painful reinvention. The best-case outcome of a Harley-style turnaround is: survive, recover partially, operate in a more competitive market with lower margins. Forever.
That is the optimistic scenario for Hollywood. A real response will require more than Scenario Two tinkering. It will require decisive leadership from visionary executives who are willing to commit to AI at the core of the production process and drive that commitment through to success, accepting that the first generation of results may be imperfect and that the journey will be uncomfortable. These executives must be sheltered from the blowback from the chorus of moral outrage (and the occasional swirly-obsessed hellion). The incentive structure inside studios that punishes failure and does not reward courage must be amended to promote risk-taking and boldness.
Otherwise, if the media companies settle on the alternative approach, which consists of a mediocre strategy of adopting the minimum viable AI at the margins of the process with feeble executive support, then they will almost certainly ensure that the disruption will arrive on schedule. If that happens, the major motion picture companies will be unprepared for it.
The studios that figure this out first will seize a larger share of the smaller world that follows. Those that spend the next three years optimizing VFX budgets while leaving the assembly line intact will discover to their dismay, as Harley did in 1980, that the competitor was not building a better version of what they already had. They are building the next generation customer.
Coda: See you at The Lot That Tom Ince Built
On Wednesday this week, I will join more than 1,200 filmmakers, developers, and tech entrepreneurs at AI on the Lot.
The conference will be held at the Culver Studios in Culver City, where Amazon Studios is the anchor tenant. The lot was built in 1918 by Thomas Ince, the silent-film-era producer who pioneered the assembly-line method of filmmaking that has governed Hollywood motion picture production for the past century.
On the lot where Ince’s studio once stood, a new industry is convening to figure out how to dismantle everything he invented.
+++++++++++++
For my readers outside the film and TV industry, I encourage you to draw a parallel between the innovator’s dilemma as faced by the major studios and the situation that faces the incumbents in your own industry. I am keen to learn from your perspective.
Rob Tercek is the author of Vaporized, which was distinguished as the International Book of the Year at the Frankfurt Book Fair in 2016. He co-founded Nura Studios AI and is the co-host of The Futurists podcast.
The Honda/Harley example can be seen in other business industries over the years, this is an excellent article that should spur deep contemplation for everyone
Hard to ignore the comparison with Harley Davidson. Great analysis and discussion.