The higher the two-qubit fidelity, the lower the overhead for quantum error correction. This is especially preferable due to the fact that cheap QRAM would be useful for practical applications, QRAM is expensive due to error correction, and classical RAM doesn't need error correction.
Is there an upper bound on how much two-qubit fidelities can be improved? Is there some type of no-go theorem that says, for example, "The highest 2 qubit fidelity you can get from transmons is X?". If so, what is it? The highest prediction I could find is $10^{-6}$ for Majorana qubits. I'm not aware of a lower prediction.
