The $50 Film
Who Gets to Make Movies When AI Rewrites the Rules
Tim, a well-known film and video blogger on Huxiu, one of China’s leading tech media platforms, had spent years evaluating creative tools. He had seen plenty of impressive software. But when he finished testing Seedance 2.0, ByteDance’s latest AI video generation model, something shifted. The model had taken a single photograph of him and, without any audio samples whatsoever, generated a video featuring what appeared to be his voice -- his speaking rhythm, his pauses, his cadence. It reconstructed his studio space from a partial screenshot, filling in the unseen half of the room with near-photographic accuracy. No voice recordings had been provided. No 3D scan of his workspace. Just one photo. “For the first time,” Tim wrote, “I felt fear toward AI.” Within days, his reaction rippled across Bilibili and Xiaohongshu, China’s major content platforms, where film editors, animators, and voice actors echoed the same anxiety: their livelihoods were suddenly, viscerally at stake.
Half a world away, at the Short Shorts Film Festival & Asia international conference in Tokyo, a Senegalese filmmaker named Hussein Dembel Sow offered a starkly different response to the same technological wave. “Where I come from, we are happy,” he told the assembled filmmakers from ten countries, “because it will help us build new industry.” Sow is currently developing a generative AI film studio in Africa, aimed at democratizing audiovisual creation on a continent where science fiction and fantasy films have long been out of reach due to the prohibitive cost of visual effects.
The same AI video tools. Fear on one side, hope on the other. The reality, as it turns out, is more complicated than either reaction suggests. What exactly are these tools democratizing? And what new boundaries are they drawing?
A One-Minute Film for $50
The numbers coming out of China’s AI micro-drama market answer the first question with startling clarity. The cost of producing a comedy-style AI animated short -- the kind featuring exaggerated emoji-like characters in slapstick scenarios -- has fallen to between 50 and 150 yuan per minute, roughly seven to twenty US dollars. A team of six people can complete an entire short drama in three days. At the high end, more polished AI-animated dramas cost 1,000 to 3,000 yuan per minute, with teams of around fifteen working over four weeks. Compare this with traditional animation, where per-minute production costs typically run into thousands or even tens of thousands of yuan, and the shift becomes clear: an order-of-magnitude cost collapse has occurred in the span of months.
(The above is a Spring Festival greeting video created by the Global South Academic Forum using Seedance 2.0.)
This is not a laboratory curiosity. It is a functioning market generating real revenue. On Douyin, China’s version of TikTok, the number of fully AI-generated short dramas in the platform’s top 5,000 rankings surged from just 4 in January 2025 to 217 by November -- a 54-fold increase in under a year. China’s AI-animated micro-drama market reached an estimated 189.8 billion yuan (approximately $26 billion) in 2025, growing 276.3 percent year-on-year according to market research firm iiMedia. Some AI short dramas are reporting net profit margins above 50 percent. On Douyin alone, original AI-animated drama uploads exceeded 60,000 titles in 2025, up from fewer than a thousand the year before.
The answer to “who gets to make a film” is being rewritten. Where filmmaking once required expensive cameras, lighting rigs, sound studios, and teams of specialized technicians, the barrier is shifting from capital and equipment to creativity and coordination. A person with a compelling story and basic digital literacy can now produce a watchable short drama for the cost of a restaurant meal. The tools, at least, have been democratized.
The Reality of 30,000 Prompts
But that is only half the story. If AI tools truly let anyone make a film, why did Xu Dapeng, a director specializing in AI short dramas, report that after interviewing more than 200 candidates, fewer than 10 were actually suited for AI filmmaking?
The answer lies in what “making an AI film” actually involves. German filmmaker Marcel Barsotti, who produced the AI film IMPERIA, revealed at the SSFF&ASIA conference that his production required layering more than 30,000 individual prompts. Thirty thousand. Each one a precise instruction to the model about framing, lighting, movement, expression, pacing, and continuity. The popular fantasy of “one-click filmmaking” -- type a sentence, get a movie -- is exactly that: a fantasy. What AI filmmaking actually demands is a new compound skill set that blends prompt engineering, directorial vision, and technical coordination. The tools have become accessible. The ability to use them well has not.
The distinction matters because it shapes who benefits and who gets squeezed. Adobe Senior Product Marketing Manager Kylee Pena offered a measured assessment of AI’s role in professional production: it is about “eliminating the friction that interrupts the creative flow,” she explained, allowing editors to “spend less time on tedious tasks and more time shaping story, pacing, and tone.” From this vantage point, AI is a liberating force -- freeing creators from drudgery to focus on craft.
But in the trenches of China’s AI micro-drama industry, the lived experience tells a different story. The efficiency gains that AI enables have not translated into better conditions for the people wielding the tools. Production fees for outsourced AI animation have plummeted from 3,000 to 5,000 yuan per minute in 2024 to just 500 to 1,000 yuan per minute -- in some cases, as low as 400 yuan, with desperate freelancers undercutting each other to survive. Working until two or three in the morning has become routine. One frontline creator captured the atmosphere with bleak precision: “The boss always thinks that since you’re using AI, it should be fast, good, and cheap -- that’s just expected now.” The technology is not replacing people outright. It is compressing and reshaping their roles, often squeezing their incomes in the process.
There is a structural dimension to this compression that extends beyond individual creators. In an era when “anyone can make a film,” the distribution channels still belong to platforms. In China, companies like ByteDance (Douyin), Kuaishou, and Hongguo simultaneously serve as AI tool providers, content distributors, and rule-makers -- players and referees in the same game. Runway CEO Cristobal Valenzuela has declared that “AI is basically eating Hollywood,” but Runway itself is pivoting from a tool company into a platform company, offering APIs for developers to integrate video generation into their own applications. The pattern is not unique to any one country. It is global: as production costs crash toward zero, the companies that control distribution infrastructure gain disproportionate structural power. In China’s AI micro-drama market, the return on investment for paid dramas has already fallen to 1.1 -- meaning that for most creators, only content selected and promoted by platforms can turn a profit.
If AI is a double-edged sword for Chinese creators -- lowering barriers while intensifying competition and platform dependence -- then what does it mean for the Global South? In places where even traditional filmmaking equipment has been scarce, an imperfect democratization might still represent a transformative leap in capability.
Senegal’s Film Dream
Return to Hussein Dembel Sow, the Senegalese filmmaker whose optimism opened this article. At the SSFF&ASIA conference, he went beyond expressing general hope. He made a specific, structural argument: “Without AI, genres like fantasy or science fiction that require large-scale VFX will forever remain out of reach.” For filmmakers across much of Africa, this is not hyperbole. It is a statement of economic and material reality -- a description of genres that require tens of millions of dollars in post-production budgets that simply do not exist. Sow is building a generative AI film studio on the continent precisely to address this gap -- to make it possible for African creators to tell stories that have traditionally required budgets they could never access. Shin Chul, the director of South Korea’s BIFAN International Fantastic Film Festival, reinforced the point from another angle: “AI allows young creators to start from the same starting point as James Cameron.”
The data behind this aspiration is both inspiring and sobering. Nigeria’s Nollywood produces approximately 2,500 films per year, making it the world’s second-largest film industry by volume. But budgets typically fall below $50,000 per film. African cinema as a whole accounts for less than 2 percent of the global box office. The continent’s share of global investment in AI-powered audiovisual technologies is similarly marginal -- under 2 percent. Yet the appetite for change is real: nearly 80 percent of African creators under the age of 30 say they are eager to use AI in their creative work.
Chinese AI video tools offer these creators a concrete, affordable option. Kling, developed by Kuaishou, generates a ten-second video clip for approximately $0.50. Google’s Veo, one of the leading Western alternatives, costs around $2.50 for the same output -- five times as much. Alibaba’s Wan model is fully open-source, and its 1.3-billion-parameter version requires only 8 gigabytes of video memory to run locally. That means a creator with a mid-range laptop can generate AI video without paying for expensive cloud computing services. When a minute of comedy-style AI animation costs between $7 and $20 to produce, the financial barrier for a filmmaker in Lagos, Dakar, or Nairobi drops to a level that was unimaginable just two years ago.
But leapfrogging is not automatic. A report from the Center for Strategic and International Studies notes that while AI is expected to add $19.9 trillion to the global economy by 2030, only 3 percent of those economic gains are projected to flow to Latin America, and just 8 percent to Africa and other developing regions. The infrastructure gap is stark: in sub-Saharan Africa, only one in four people has reliable internet access. And the risk of what researchers call “digital colonialism” -- excessive dependence on foreign AI platforms that erodes data sovereignty and cultural autonomy -- is not theoretical. Congolese producer Leonard Bassegela put it plainly: “If we don’t build our own tools, our stories will be told by algorithms trained somewhere else.”
The choice taking shape for Global South creators is becoming clearer. AI tools -- especially low-cost Chinese tools and open-source models -- genuinely offer the possibility of vaulting over traditional development stages in filmmaking. But whether that leap actually happens depends on something deeper than tool access: it depends on autonomy over both the tools and the narratives they produce.
Who Controls the Story
The real question is no longer whether AI can make a film. It can, and for as little as $7 a minute. The question is what determines whose film gets seen once AI makes it possible for everyone to create one. The answer is shifting from “who has the equipment and the funding” to a new set of gatekeepers: who masters the emerging compound skills of AI-assisted creation, who controls the platforms that distribute content, and who retains the capacity to tell their own stories in their own voice.
For Global South creators, the most critical task is not acquiring cheap tools -- those already exist. It is investing in the new hybrid skills that AI filmmaking demands, building local capacity for AI model training and adaptation, and ensuring that distribution channels are not entirely controlled by foreign platforms. The tools are good -- genuinely, materially good. But what matters most is who controls the tools, and who controls the story.
Tim’s fear and Hussein Dembel Sow’s hope are both grounded in reality. AI video tools are genuinely liberating creators from constraints that once made filmmaking the province of the well-funded few. They are also genuinely redistributing power -- toward platforms, toward those who master new skills, and potentially toward those who control the underlying technology. The open question -- one that will be answered not by technology but by investment, policy, and collective action -- is whether, in this redistribution, the Global South will be a participant or a spectator.