The Pokemon Company is launching a new competition that challenges developers to build artificial intelligence agents for the Trading Card Game. Organizers partnered with Kaggle, Google, and Nvidia to create this event, which tests how well machine learning models handle complex deck-building strategies. Participants must design AI systems capable of navigating the strategic depth of card-based combat.
Developers build artificial intelligence agents to compete in card-based combat
The competition splits into two distinct tracks that run concurrently throughout the summer. The Simulation Category runs from June 16 through August 17, 2026, focusing on direct match performance. The Strategy Category extends until September 14, 2026, evaluating broader tactical decisions and deck construction over longer periods.
Developers entering the Strategy track compete for a $50,000 prize pool alongside Google Cloud credits for their work. The top-performing teams in this category will advance to finals later in the year, marking the conclusion of the challenge. This structure rewards both immediate gameplay efficiency and long-term strategic planning within the digital card environment.
The event highlights growing interest in applying machine learning to structured games with hidden information and complex rule sets. AI agents must learn to predict opponent moves while managing limited resources across multiple turns. The competition provides a concrete benchmark for testing these capabilities against established game mechanics.
Source: Pokemon TCG, DEXERTO
