The deadline is for migration, not for building a quantum computer. That’s the boring part, and the expensive part.

The part people keep missing is that quantum is not a general-purpose break-the-internet switch, it targets the stuff we lean on for identity and key exchange, mostly RSA and ECC in TLS, code signing, software updates, and long-lived records. Once Shor is practical, the ugly scenario is not just live traffic getting popped, it is “harvest now, decrypt later” against anything with a long confidentiality tail, which is why archives and government data care so much. That is also why post-quantum crypto feels weirdly unlike the rest of the quantum stack, because the deadline is driven by migration time, not by when a machine with enough qubits shows up. Replacing algorithms in the field means inventories, cert chains, firmware, HSMs, protocol upgrades, and all the boring compatibility glue, and that tends to eat years (sometimes more than the crypto itself).