[AI multi-modal wave is coming! ChatGPT is expected to meet the demand for heavy upgrade computing power.] The latest ChatGPT can tell stories, conduct voice conversations, and read pictures; OpenAI calls the image-seeing model GPT-4V(ision), and has tested GPT-4V's captCHA cracking and geolocation capabilities. The size of voice and image data is significantly higher than that of text, and brokers point out that the training and reasoning computing power needs of multi-modal large models will rise sharply.