Chinese Cities Court AI Filmmakers With Subsidies and Computing Power

Chinese Cities Court AI Filmmakers With Subsidies and Computing Power

Local governments in China are applying the familiar industrial-policy playbook — subsidies, tech clusters and computing power — to AI-generated film, just as production costs have plummeted and platforms flood with content that rarely finds an audience. According to the analytics firm DataEye, 221,900 new AI series launched on Douyin in the first half of the year, but only 1,055 passed 100 million views.

When Zhu Zhili was looking for a base for an AI film studio in China two years ago, there was only one obvious choice: the southern metropolis of Shenzhen, with its vast technology ecosystem. Today the picture is different. According to Reuters (Ellen Zhang and Alessandro Diviggiano, September 23, 2026), Zhu, who heads the AIGC department at the production company China Wit Media, is now contacted daily by cities that want him.

"I'm contacted every day by a range of cities, from big metropolitan areas to smaller places," Zhu said, according to Reuters, "in the hope that we can establish either technology or the company there."

The story Zhu tells is the core of what Reuters describes as a nationwide rush: China is trying to turn AI-generated video into an industry, using the same tools it deployed to build electric vehicles, solar panels and robotics into globally competitive sectors.

The industrial-policy playbook

Reuters draws a pattern recognizable from earlier Chinese state bets: local governments are promising tech clusters and subsidies to AI filmmakers and startups. The goal is to reproduce the concentration of capital, talent and suppliers that turned sectors like EVs and solar into export industries — this time with AI-generated content as the raw material.

The concrete measures sit with the big cities. Shanghai introduced measures in May to accelerate production of AI-driven micro-dramas, according to Reuters: producers are offered computing power and cloud-based AI models to cut costs, along with support for overseas distribution. That last point is worth noting — the city is not only betting on getting production to happen locally, but also on opening foreign markets for the result.

Beijing, according to Reuters, has a 260 million yuan fund to support audiovisual technology. The city's Huairou district, a hub of China's film industry, offers producers of short AI dramas vouchers that reduce computing costs. The scheme thus targets one of the main technical cost lines in AI video production head-on: GPU compute.

Shenzhen, the starting point of Zhu's story, is cited by Reuters as the city with the big technology ecosystem — the details of today's Shenzhen offering are not covered in the Reuters report.

Costs are collapsing

The subsidy push is hitting an industry whose economics are already shifting. According to the broadcaster CCTV, the cost of producing AI-driven short dramas fell from 5,000 yuan ($747) per minute to just a few hundred yuan in the first half of 2026. The figure is presented as an industry average, with no known methodology behind it, and it comes from a state broadcaster. But the direction is unambiguous, and it is corroborated at the individual level.

One of those individuals is Pan Xiaojun, a postgraduate film-directing student based in Hainan in southern China. "An ordinary wedding scene can cost me 60,000 yuan to shoot conventionally," Pan said, according to Reuters. "But with AI, I can bring the same surrealistic scene to life for just 1,400 yuan."

That is a reduction of more than 97 percent on a single scene type, and it illustrates the mechanism behind the whole push: when marginal production costs approach zero, the barrier to making content effectively disappears. That is what makes AI film a target for industrial policy — and simultaneously creates the problem below.

A flood on the supply side, a drought on the demand side

The other side of the story is that the content largely fails to find an audience. According to DataEye, as reported by Reuters, 221,900 new AI series launched on Douyin — the Chinese version of TikTok — in the first half of the year. Only 1,055 of them passed 100 million views, described as a typical benchmark of success.

That means roughly 0.5 percent of launches reached the success threshold. It is too early to declare overcapacity: the figure rests on a single Douyin statistic, and the 100-million threshold is DataEye's and Reuters's framing, not an objective measure. In an industry where production costs have fallen as sharply as the CCTV figure suggests, an explosion in volume is in any case to be expected. But the combination — subsidized entry of new production capacity, collapsing production costs and a vanishingly small share of hits — resembles the trajectory of earlier Chinese industrial bets in phases where capacity grew faster than demand.

What is not verified

Most of this comes via a single source: a Reuters article dated Beijing/Shenzhen, September 23, 2026, distributed via MSN. The underlying documents — the municipal measures from Shanghai and Beijing/Huairou, the DataEye report and the CCTV figures — have not been made available for independent review, which makes the precision of the details uncertain, even though the reporting is internally consistent and dated.

It is also worth keeping in mind that the CCTV figure refers to average cost per minute, while Pan's anecdote concerns a single scene — two different units of measurement pointing in the same direction, but which cannot be merged into one precise number.

What is worth watching

The most important indicator going forward is whether subsidy policy actually tracks demand, or whether it helps build even more capacity in a market where 221,900 launches are already competing for attention. The Huairou vouchers are an interesting case: they lower computing costs, which are a real cost for producers, and can therefore be justified as industry support regardless of where the content ends up.

Zhu's experience of being contacted daily suggests that intercity competition for the AI film industry is already underway. If it follows the pattern of EVs and solar, the result could be a concentration of capacity in a few clusters — and a risk that supply keeps growing faster than the number of viewers.

Sources: Reuters (Ellen Zhang and Alessandro Diviggiano), "China fuels rush to turn AI video into an industry," September 23, 2026, via MSN.