1.6K
AIA report from MIT Technology Review highlights just how energy-intensive AI video generation can be. Researchers estimate that creating a 5-second AI video may require around 3.4 million joules of energy, roughly comparable to running a microwave for more than an hour.
The reason is that video models process enormous amounts of data. Unlike text or images, AI video systems must generate thousands of frames while calculating motion, lighting, and visual details across millions of pixels.
By comparison, a typical AI text response may consume between about 114 and 6,700 joules, while AI image generation can require hundreds of times less energy than video generation.
Researchers, including Sasha Luccioni from Hugging Face, analyzed open-source models and found that video’s pixel-heavy structure dramatically increases computing demands.
As tools like Sora and other AI video platforms grow in popularity, experts say the electricity demand of data centers could rise significantly.
The findings highlight a growing question in the AI world: how to balance powerful new capabilities with their environmental impact.
Follow for more AI insights, tech trends, and future technology updates.
#ai #artificialintelligence #aivideo #technews #futureofai
@ai.momentum.inc
