The technology that killed Hollywood...for the fourth time

Posted on: 6 April 2026

Every decade or so, someone explains to you that Hollywood is finished. The mechanism is precise, almost ritualistic: a new technology emerges, a group of very intelligent people produces very convincing analysis about how cinema will never be the same again, and then cinema remains substantially the same with a few more special effects.

The first time I paid close attention to this was with pure CGI in the nineties. When Final Fantasy: The Spirits Within arrived in 2001 with entirely computer-generated actors, it seemed like definitive proof that human presence on screen had become optional. The film cost $137 million, earned $85 million, and took Square Pictures down with it. The actors were still there.

Then came Robert Zemeckis with Beowulf in 2007 and the ambition escalated: not invented characters but digital clones of real actors. Anthony Hopkins, Angelina Jolie, Ray Winstone recreated frame by frame through motion capture. The technology was remarkable. The result was disturbing in the precise sense: the characters were similar enough to human to generate discomfort without being human enough to convince. Perception researchers call this the uncanny valley, the zone where something almost-human produces repulsion rather than empathy. The film barely recouped its budget and Zemeckis abandoned that road entirely. Nobody serious has gone back down it since.

Now it is generative AI and Bollywood.

The Reuters piece published yesterday documents with journalistic precision what is happening in the studios of Bengaluru: production costs down to a fifth, timelines compressed to a quarter, entire mythological films generated without physical sets. The numbers are real. The technology works. And the market responds, at least initially.

The case worth examining closely is Raanjhanaa. A 2013 film, lead actor Dhanush, tragic ending: the protagonist dies. Last year Eros Media released a version with the finale altered by AI, the protagonist opens his eyes, his lover smiles through tears. Thirty-five percent of available tickets sold in the release month. Dhanush protested publicly, saying they had stripped the film of its soul. The AI Mahabharat series produced with the same technology sits at 1.4 out of 10 on IMDb.

The market bought the ticket out of curiosity. It did not buy the film. The same dynamic as Beowulf, the same uncanny valley, the same unease that audiences struggle to name but recognise immediately when they feel it.

This is the distinction that technology vendors tend not to make, or choose not to make, because conflating it serves their interests. The media pressure that accompanies every technological disruption in cinema follows a consistent template: urgency is manufactured, anyone not adopting the new tool immediately is framed as a condemned dinosaur, efficiency metrics are presented as creative destiny. It is a sales strategy dressed as industrial analysis. The British entertainment industry has been on the receiving end of this pitch several times now, from digital projection to streaming to virtual production. The structure of the argument never changes.

The underlying mechanism does not change either, because audiences do not buy images. They buy relationship with a specific presence. Shah Rukh Khan is worth what he is worth not because of how he acts but because of what he has become in the minds of hundreds of millions of people over decades. That function is not technical and has no technical solution. You can generate a perfect face, synchronise lips across twenty-two languages, eliminate every physical imperfection. You cannot generate the history that a specific body carries with it, the accumulated expectations and memories that transform an actor into a cultural symbol.

Bollywood understands its own business, that much is clear. It knows how to work within constrained budgets, how to build fragmented linguistic markets, how to produce at volume. AI will give it real tools to do better what it already does, particularly in dubbing across India's twenty-two official languages, which is the most credible use case in the entire Reuters piece. But Hollywood is Hollywood for reasons that have nothing to do with production costs.

The short memory is the actual protagonist of this story. Each generation discovers for the first time that technology might make actors redundant and reacts as though it were news. It is not news. It is the same film, with different special effects.