Heads-up no-limit Texas Hold’em is a primary benchmark challenge for AI. Due to the hidden information in poker, techniques used for games like chess and Go are ineffective. We present Libratus, the first AI to defeat top human professionals in no-limit poker. Libratus features three main components: pre-computing a solution to an abstraction of the game which provides a high-level blueprint for how the AI should play, a new nested subgame-solving algorithm which repeatedly calculates a more detailed strategy as play progresses, and a self-improving module which augments the pre-computed blueprint based on opponent behavior.