Spatial AI is based on computation over spatial domains. Calculations suggest that, due to what can be named “the miracle of spatial information”, the number of spatial questions required to identify the location of a square cell on it tends to zero as the size of a square map increases. Yet, physics sets an upper bound (1093 bits) to any kind of computing on the face of the earth (the Bremermann limit) and this limit applies irrespective of the stage of development of AI. So the expansion of big spatial data and spatial combinatorial explosions prompt us to examine possible limits to the capabilities of Spatial AI. While the Landauer limit might be overcome by Quantum AI, avoiding the obstacle of transcomputation in spatial data processing seems more challenging. However, as calculated here, classical computing implies that the total number of bits required to process all possible square binary map configurations does not exceed the size 17 × 17, while for maps 10 × 10 the calculation of all possible configurations will be possible by classic computation for only up to 8 colors.