LLeMpower: Understanding Disparities in the Control and Access of Large Language Models