---
title: "A ChatGPT prompt equals about 5.1 seconds of Netflix"
description: "!https://simonwillison.net/2025/Nov/29/chatgpt-netflix"
date: 2025-12-02
published: true
tags:
  - ai
  - thought
template: link
---


<div class="embed-card embed-card-external">
  <a href="https://simonwillison.net/2025/Nov/29/chatgpt-netflix" class="embed-card-link" target="_blank" rel="noopener noreferrer">
    <div class="embed-card-content">
      <div class="embed-card-title">A ChatGPT prompt equals about 5.1 seconds of Netflix</div>
      <div class="embed-card-description">In June 2025 Sam Altman claimed about ChatGPT that &#34;the average query uses about 0.34 watt-hours&#34;. In March 2020 George Kamiya of the International Energy Agency estimated that &#34;streaming a …</div>
      <div class="embed-card-meta">Simon Willison’s Weblog &middot; simonwillison.net</div>
    </div>
  </a>
</div>


This feels very promising for the future as we enter a world that is more and more dependent on AI that inference is so cheap.  I did not understand the scale to how much cheaper inference is compared to training.  As we get better with training I imagine this gets significantly better as well.  I know they all claim to be profitable on inference, but scrolling through Simon's feed here you see several articles on the stark difference.