Current commonly used public key cryptography systems are based on the hardness assumption of factorization and/or discrete lograrithm.
Both these problems are solved efficiently using Shor's algorithm using a quantum computer.
Should someone build a quantum computer capable of running the algorithm with thousands of qbits and the ability to apply enough operations on them without decoherence, then it would break RSA and diffie-hellman including elliptic curvre variants.
This breaks all commonly used public key cryptography but does not break all known public key cryptography. There are other asymmetirc algorithms, for example those based on lattice problems which are currently believed to be secure even in the face of quantum computers.