A feature-length film produced in two weeks for less than $500,000 just screened at the Cannes Film Festival. Most of that budget didn’t go to actors, sets, or catering trucks. It went to GPUs.
Higgsfield AI, a San Francisco-based generative video company, premiered Hell Grind at Cinema Olympia in Cannes on May 21. The 95-minute sci-fi heist film is being positioned as the first complete AI-native feature to demonstrate that long-form narrative filmmaking, the kind with consistent characters, coherent world-building, and an actual plot, can be done entirely with generative AI tools.
The numbers that matter
A traditional film production of comparable length and genre typically runs around $50 million. Hell Grind came in under $500K. That’s roughly a 99% cost reduction.
About 80% of that budget went directly to AI compute costs. The production team consisted of just 15 people, including directors and editors who used Higgsfield’s proprietary suite of AI tools: Dreamina-Seedance 2.0, Soul Cinema, and Soul Cast.
The film was directed by Aitore Zholdaskali and co-written with Adilkhan Yerzhanov, a two-time Cannes Official Programme filmmaker.
From start to finish, the entire production took 14 days.
How Higgsfield got here
Higgsfield had been testing its technology on progressively longer formats before attempting a full feature. The company previously produced a 22-minute pilot and shorter series projects called Arena Zero and Zephyr, essentially using them as proving grounds for the character consistency and narrative coherence problems that have plagued AI video generation.
An industry preview and trailer launch preceded the main screening, held on May 16 during a special event co-hosted with Goldfinch, a film investment and production company.
What this means for investors and the broader market
Higgsfield AI has no blockchain integrations, no token, and no apparent interest in Web3 distribution models. Some low-cap meme tokens on Solana share similar names, but they have zero connection to the company or the film.
For AI infrastructure companies, the fact that compute costs consumed 80% of Hell Grind‘s budget is a signal worth paying attention to. As AI filmmaking scales, demand for GPU compute will scale with it.