The usual framing of quantum risk puts the deadline in the future: when a cryptographically relevant quantum computer exists, today's public-key cryptography stops protecting anything. That framing is comfortable and it is wrong for a large class of data.
Capture is cheap, patience is free
An adversary who can intercept encrypted traffic does not need to decrypt it now. Storage is inexpensive and indefinite. Traffic captured this year can sit untouched until the capability to break it arrives, at which point it decrypts retroactively — including the sessions you considered safe when they happened.
The relevant question is not when the machine arrives. It is how long your data must stay confidential after today.
Which data this actually applies to
Not everything needs to move first. A session token that expires in an hour is uninteresting to a decade-long capture programme. The exposure concentrates in data whose confidentiality requirement outlives the algorithm protecting it:
- Health and genomic records, protected for a subject's lifetime
- State and defence material under multi-decade classification
- Intellectual property, formulations, and design data
- Long-lived credentials and root key material
- Financial and legal records with statutory retention periods
This is why discovery comes first. You cannot rank exposure you have not inventoried, and an unranked migration programme defaults to migrating whatever is easiest — which is rarely what matters.