Yeah, except none of that is true. An RTX 4060 can generate a Stable Diffusion XL image in 10 seconds.
A 4060 consumes 110 Watts. That's 1100 Joules. That's 0.0003055556 KWH.
The 1.35 kWh per 1000 number that is often (mis)quoted is for Multi-Modal models, which are rarely used just for generating images.
To be clear. It uses the same amount of energy to play a video game for one hour as it does to generate three hundred sixty images.