Today’s advanced AI models require exponentially increasing memory capacity and bandwidth, as Large Language Models (LLM) including GPT-4 used for ChatGPT and Recommendation Engines are memory bound rather than compute bound. Celestial AI is here to solve that problem.