'Thermodynamic computing' could slash energy use of AI image generation by a factor of ten billion, study claims — prototypes show promise but huge task required to create hardware that can rival current models

A new study suggests that 'thermodynamic computing' could significantly decrease the energy consumption of AI image generation by a factor of ten billion compared to current tools. While prototypes show promise, developing hardware to match existing models remains a significant challenge. Researchers at Lawrence Berkeley National Laboratory have demonstrated the potential for thermodynamic computing to generate images with much lower energy costs than traditional digital hardware. Although still in its early stages, this research could pave the way for more energy-efficient AI image generation in the future.