According to the paper: > All experiments were run using third generation Google Cloud TPUs [12]. For each board game, we used 16 TPUs for training and 1000 TP

According to the paper: > All experiments were run using third generation Google Cloud TPUs [12]. For each board game, we used 16 TPUs for training and 1000 TPUs for selfplay. For each game in Atari, we used 8 TPUs for training and 32 TPUs for selfplay. The much smaller proportion of TPUs used for selfplay in Atari is due to the smaller number of simulations per move (50 instead of 800) and the smaller size of the dynamics function compared to the representation function.