NOVASTACKSystem Sdn Bhd
Post-Quantum Readiness
Live Proof — Not a Slide

Post-quantum TLS, running in production.

One server. One certificate. Two listeners — one classical, one quantum-safe. This page tells you which one you are on right now, tests your own browser, and hands you the commands to verify every claim yourself.

Measuring this connection

Checking…

Reading the listener that served this page.

Endpoint
Listener
Key exchange
Certificate
Let's Encrypt (ECDSA)
Load over post-quantum Load over classical

Same URL, same certificate, same machine — only the listener changes. Flip between the two and watch the verdict above change with it.

Section 01

Is your browser quantum-safe?

Listener B accepts only post-quantum key exchange. Any client that cannot do ML-KEM is refused during the TLS handshake — there is no fallback to fail into. If the request below succeeds, your browser genuinely negotiated post-quantum cryptography.

Testing your browser…

Opening a TLS connection to the post-quantum-only listener.

Section 02

What is actually deployed

Both listeners present the identical Let's Encrypt certificate for the identical hostname on the identical host. The only difference is the key-exchange group each will accept.

 Listener AListener BClassical (control)
PurposeServes this site, quantum-safeLive browser capability testEverything a normal server does today
Endpointpqc.app2u.click:9443pqc.app2u.click:9444pqc.app2u.click:443
Key exchangeX25519MLKEM768
+ classical fallback
X25519MLKEM768
only
X25519 / P-256
Quantum-safeYESYES — enforcedNO
Harvest-now-decrypt-laterProtectedProtectedExposed
TLS stackGo 1.25 crypto/tlsGo 1.25 crypto/tlsOpenSSL 3.0

Why two stacks. ML-KEM landed in OpenSSL 3.5. The classical listener runs the distribution's OpenSSL 3.0, where the algorithm does not exist at all — it cannot be switched on, only replaced. The quantum-safe listeners run a Go 1.25 TLS stack alongside it, on their own ports, touching nothing the classical server owns. That is the migration pattern this demo exists to show: add capability beside production, prove it, then move traffic — never rip out a running cryptographic stack and hope.

Section 03

Verify it yourself

Do not take our word for it. These commands run against the public internet from your own machine. You need OpenSSL 3.5 or newer — check with openssl version.

Test 1 — post-quantum listener accepts a PQC-only client
# offer ONLY the post-quantum group, no fallback allowed
openssl s_client -connect pqc.app2u.click:9444 \
  -tls1_3 -groups X25519MLKEM768 </dev/null

expected → Negotiated TLS1.3 group: X25519MLKEM768  handshake succeeds
Test 2 — the same client is refused by the classical listener
# identical client, identical hostname, classical port
openssl s_client -connect pqc.app2u.click:443 \
  -tls1_3 -groups X25519MLKEM768 </dev/null

expected → tls alert handshake failure (alert 40)  no shared group
Test 3 — a classical client is refused by the PQC-only listener
# the enforcement runs both ways
openssl s_client -connect pqc.app2u.click:9444 \
  -tls1_3 -groups X25519 </dev/null

expected → tls alert handshake failure  classical is not accepted here

Two opposite results, same server, same certificate, same minute. That is the difference between a claim and a proof.

Section 04

Why this matters, in plain terms

Quantum computers do not need to exist yet to put your data at risk. That is the part most briefings get wrong.

The attack

Harvest now, decrypt later

An adversary records your encrypted traffic today and simply stores it. When a cryptographically relevant quantum computer arrives, they decrypt the archive retroactively. Nothing you do in 2032 protects what was captured in 2026.

The exposure

Your data outlives the algorithm

Ask one question: how long must this stay confidential? Payroll, medical records, identity documents, contracts, national data — if the answer is longer than about eight years, today's key exchange is already the wrong tool.

The fix

Hybrid, not replacement

X25519MLKEM768 runs the classical and post-quantum algorithms together. An attacker must break both. If ML-KEM is ever weakened, X25519 still holds — so adopting it early carries no downside risk.

Section 05

Where the deadlines actually are

Dates published by standards bodies and regulators, not vendor marketing.

SourceInstrumentWhat it says
NACSA / Ministry of Digital
Malaysia
National PQC Readiness Roadmap
29 Oct 2025
A 2025–2030 framework across technical readiness, strategic alignment and workforce. Explicitly non-binding — each entity defines its own migration path and cryptographic-agility posture.
NIST
United States
IR 8547 — transition guidance RSA-2048 and ECC P-256 deprecated after 2030, disallowed after 2035.
NSA
United States
CNSA 2.0 ML-KEM-1024 and ML-DSA-87 mandatory across national security systems by 2035, with networking equipment moving far earlier.
Bank Negara Malaysia RMiT Does not mandate PQC. It does harden cryptographic key management and algorithm governance — the crypto-agility groundwork a migration depends on.

An honest note on urgency. No cryptographically relevant quantum computer exists today. Expert surveys put roughly a one-in-four chance by 2030 and one-in-two by 2035. Anyone quoting you a precise date is selling fear. The reason to move now is not the date — it is that harvest-now-decrypt-later makes the exposure retroactive, and that migrating a real estate of systems takes years of planning, not a weekend.

Section 06

How we approach a migration

The order matters more than the algorithm.

  1. Cryptographic inventory first. Which system uses which algorithm, for which data, with what confidentiality lifetime. You cannot migrate what you have not mapped — and this is precisely what the NACSA roadmap asks each entity to produce.
  2. Rank by data lifetime, not by system importance. A low-traffic archive holding thirty-year records outranks a busy service holding session data.
  3. Deploy hybrid key exchange beside production. Exactly what this page demonstrates — new capability on its own listener, proven under real traffic, before anything is switched over.
  4. Leave signatures for later, deliberately. Certificate signatures are not yet available from public CAs and the threat is not retroactive. Key exchange is where the retroactive risk lives, so key exchange goes first.
  5. Build for agility, not for one algorithm. The goal is an estate where the next algorithm change is a configuration decision instead of a rebuild.