LLMs do use more water than image generators, but guess what? They used those too!
The company used several Large Language Models to generate the creative. (Two other versions of the “holidays are coming” ad were also made this year, an AI one with another studio, Silverside, and a traditionally shot one without the stylized animals, for some Latin markets.)
Source: Coca-Cola Is Trying Another AI Holiday Ad. Executives Say This Time Is Different (The Hollywood Reporter - November 3, 2025)
Not only that, the ad also appears to have been made using generated video.
...the ad team “churned out” more than 70,000 video clips to create a finished product, which sounds incredibly inefficient. The important part to Coke, though, was that it only took five “AI specialists” and a month to make. But in total, around 100 people were involved in the new commercial campaign, per the WSJ.
Source: Coke’s New AI-Generated Ad Required 100 Staff and 70,000 AI-Generated Clips, and It Still Looks Like Garbage (Futurism - November 5, 2025)
Generating video is a lot more intensive than image generation.
The new model uses more than 30 times more energy on each 5-second video: about 3.4 million joules, more than 700 times the energy required to generate a high-quality image. This is equivalent to riding 38 miles on an e-bike, or running a microwave for over an hour.
Source: We did the math on AI’s energy footprint. Here’s the story you haven’t heard. (MIT Technology Review - May 20, 2025)
So there's no "gotcha" here. Image generation is less wasteful than video generation or LLMs, but that's completely irrelevant.
jbbartram-illu
jacobvanloon
prguitarman
blakekathryn
izzy806
gummygunk
