OpenSSL's president says the post-quantum fix is 'architectural, not bigger pipes'

The first connection a phone makes to its bank in the morning, the signed software update pushed to a router overnight, the certificate a browser checks before loading a news site: all of them run, at some layer, through OpenSSL, the open-source cryptographic library that secures most of the public internet. Right now, every one of those connections is being rebuilt, and the work has almost nothing to do with making the network faster.

Tim Hudson, president of the OpenSSL Corporation, told TechRadar Pro ahead of OpenSSL Conference 2026 in Prague next month: "The fixes are architectural, not bigger pipes." In plain English, post-quantum cryptography is a redesign of how trust moves across the network, not a bandwidth upgrade.

The threat driving the work is the one security researchers call "harvest now, decrypt later": adversaries who cannot read today's encrypted traffic believe they will be able to once a cryptographically relevant quantum computer exists, so they are already collecting it. Some already-exfiltrated data, including state secrets, medical records, and anything with a long shelf life, is on an effective countdown. The question is no longer whether the locks change, but how, and on whose schedule.

Hudson, on the record with TechRadar, frames the cost in concrete terms. The post-quantum signature scheme known as ML-DSA-44 produces, by Hudson's characterisation, a 2,420-byte signature and a 1,312-byte public key, against the 64 bytes of today's ECDSA P-256. Substituting the new primitives naively, with a certificate chain and two certificate-transparency timestamps, adds 7KB to 10KB to every new connection. The cost shows up first at the network's edges.

The extra weight hits a specific threshold: the internet's standard initial congestion window sits around 14KB, the figure Hudson cites as the cap on how much data a connection can send before the other side acknowledges. Push past it, and fresh connections need an extra round trip. On a wired link, that delay is invisible. On a mobile handover, a satellite hop, or a constrained device waking up, it is a tail-latency problem that compounds with the millions of new connections made every minute.

If the answer is not bigger pipes, Hudson argues, it is three concrete redesigns already in motion. The first is Merkle Tree Certificates, a way of compressing a long certificate chain into a single short proof a client can verify without downloading every intermediate. The second is trust anchor negotiation, which lets a client and server agree on which root they both recognise instead of forcing the server to ship a long chain. The third is suppressing intermediates, so a server sends only the certificate the client needs.

Each of those is a change to the shape of the handshake, not to its size. The work is happening inside OpenSSL and its peers. OpenSSL Conference 2026, which ITbrief NZ reports will also cover software supply chain and digital trust, is where much of that design will be discussed in public.

Hudson names four places the transition will bind. First, published guidance is written for large regulated enterprises, and smaller operators have no analogue. Second, validation throughput, getting new primitives through FIPS and Common Criteria, is itself a binding constraint, and certified modules trail published standards by years. Third, there is no credible plan for deployed embedded and operational-technology systems with 15 to 25-year service lives, the kind of fleet that cannot be reflashed on a regulator's schedule. Fourth, procurement is still buying algorithms by checkbox, when it should be buying agility and disclosure.

That last point is where a CISO can act this quarter, Hudson argues: inventory what you have before you specify what to buy. Discovery in a complex estate takes six to twelve months, and a cryptographic bill of materials, or CBOM, is already on the regulatory agenda. The US has moved with Executive Order 14412 and OMB Memorandum M-26-15; Australia's Signals Directorate has set a refined transition plan for end-2026 and a traditional-crypto retirement date at end-2030; the EU roadmap runs through 2035 for full transition.

The shift is not a future problem. The Prague conference opens 13 October 2026, and the handshake a phone makes that morning will already be running code shaped by these decisions.