It just dawned on me that we should be calculating the number of possible public keys (not private keys) when illustrating the chances to steal someone's bitcoin. (The whole 'single atom in the universe' thing)

Since there are far fewer possible public keys than there are private keys, there is obviously a less-than 1:1 ratio between them, and therefore multiple private keys could clash and produce the same public key - and therefore would be able to spend the bitcoin associated with its UTXO set.

Am I wrong?