The Hutter Prize posits that the ability to compress data is mathematically equivalent to predicting the next token. This theoretical framework suggests that perfect compression requires a perfect model of the world. Practitioners should view LLM training as a massive compression exercise. It simplifies the goal of intelligence into a measurable data efficiency problem.