Fully PQ TLS in the WWW


Date
2025-12-03 14:35 — 15:00
Location
New Babylon Conference Centre
Anna van Buerenplein 29
2595 DA Den Haag
The Netherlands

AI-generated summary of my slides

While the industry has successfully deployed Post-Quantum (PQ) Key Exchange, the next step—PQ Authentication—presents significant infrastructure challenges due to data size and latency. The presentation proposes Merkle Tree Certificates (MTC) as a sustainable solution to enable full PQ security without breaking the web.

1. Current Status: Key Exchange is Solved

  • Adoption: Major players (Cloudflare, Chrome, Edge, Firefox) have already deployed PQ key exchange.
  • Performance: The current standard (X25519MLKEM768) adds only ~2kB to the handshake and causes a negligible ~4% slowdown.
  • Motivation: This is widely adopted because it solves the immediate “Harvest-Now-Decrypt-Later” threat.

2. The Challenge: PQ Authentication

Implementing PQ authentication (signatures) is much harder than key exchange:

  • The Bloat Problem: Replacing current authentication chains (Server cert, Intermediate cert, Handshake signature, Certificate Transparency SCTs) with PQ standards like ML-DSA adds 18–36 kB of data to the handshake.
  • Operator Hesitance: For website operators, this creates a massive performance hit for a threat that only exists in the future (the “Quantum Apocalypse”). Many are likely to “wait” rather than upgrade.
  • Infrastructure Strain: The current Certificate Transparency (CT) system is already fragile and expensive to maintain (consuming ~20TB of disk/year per log). Adding large PQ signatures would make running CT logs significantly harder and more expensive.

3. The Solution: Merkle Tree Certificates (MTC)

To solve the data bloat issue, the IETF is standardizing Merkle Tree Certificates (MTC).

  • How it Works: Instead of a traditional chain of signatures, a Certificate Authority (CA) logs the certificate into a Merkle Tree. The “certificate” becomes a proof of inclusion in that tree.
  • The Process:
    1. The CA logs the entry first, then signs a “tree head” (checkpoint).
    2. Browsers/Clients fetch these tree heads out-of-band.
    3. The server proves its certificate is valid via a short inclusion path (Merkle path).
  • Short-lived Certificates: MTCs allow for “signature-less” certificates if the client is up-to-date, or “full” certificates if not, utilizing short-lived proofs.

4. Key Benefits of MTC

  • Massive Size Reduction: MTC log entries are 97x smaller than the traditional CT approach combined with ML-DSA-44 signatures.
  • Efficiency: An inclusion proof is only ~736 bytes, compared to >7KB for multiple ML-DSA signatures.
  • Sustainability: It removes the need for intermediate certificates and makes running transparency mirrors cheaper and more reliable.

5. Timeline and Standardization

  • Standards Body: The proposal is being developed in the IETF PLANTS (PKI, Logs, And Tree Signatures) working group.
  • Backers: The initiative is being pushed primarily by Google Chrome and Cloudflare.
  • Deployment: Cloudflare and Google have announced plans to begin experimenting with MTC on the web in 2026.
Thom Wiggers
Thom Wiggers
Senior Cryptography Researcher

My research interests include (post-quantum) cryptography and protocols