Why Post-Quantum Security Can't Wait Until 2030
Most platforms say post-quantum is on the roadmap. We shipped it. Here's why harvest-now-decrypt-later attacks make waiting catastrophically expensive.
The Harvest Now, Decrypt Later Problem
Every TLS session your application uses today is being recorded by sophisticated adversaries. Not because they can break it now but because they expect to break it the moment sufficiently powerful quantum computers become available.
This strategy is called harvest now, decrypt later. And it is not theoretical.
Intelligence agencies, nation-state actors, and well-funded criminal organizations are actively archiving encrypted traffic. Your API keys, credentials, health records, and financial data are being stored in vaults waiting for the day a sufficiently large quantum computer cracks the key exchange that protected them.
When Will That Day Come?
Estimates range from 5 to 15 years. But the precise timeline is irrelevant to the threat. The data being captured today will be vulnerable on that day, regardless of when it arrives.
If you are building infrastructure that stores sensitive data with a shelf life beyond a few years healthcare, finance, government, legal you are already in scope.
What NIST FIPS 203 Gives You
The National Institute of Standards and Technology published FIPS 203 in 2024, standardizing ML-KEM-768 (formerly Kyber) as the primary post-quantum key encapsulation mechanism. This is not experimental cryptography. It is a published, peer-reviewed, standards-body-approved algorithm.
At TAS, the [lvls vault](/platform#lvls) uses ML-KEM-768 for all credential encryption. Not because it is trendy. Because the harvest-now-decrypt-later threat applies directly to the credentials that AI agents use to access production systems.
What This Means for AI Infrastructure
AI agents operate at machine speed. They pull credentials, make API calls, and access sensitive systems continuously. Every one of those credential exchanges is a potential harvest target.
A vault that relies on classical RSA or ECDH for key exchange is, over a 10-year horizon, an open vault. The keys are just not yet in the attacker's hands.
Post-quantum credential security is not a premium feature. It is the baseline for any infrastructure expected to be secure beyond the current decade.
The question is not whether to implement post-quantum security. The question is whether you waited too long.