Live monitoring Updated 72-hour alert levelPriority Editor: Frédéric LENQUETTE
CYBERIACKCybersecurity & AI intelligence · by Frédéric LENQUETTE
Post-quantumImportant

Sanofi is quantum-proofing its encryption, but its software vendors are lagging

The pharmaceutical group detailed its approach to LeMagIT: automated inventory, six priority criteria, certificate testing. A useful case study that also shows where things get stuck.

Translated from French by AI. The original French article is the reference version.

Article illustration : Sanofi is quantum-proofing its encryption, but its software vendors are lagging
AI-generated illustration

The script runs every day. Written in Python by Sanofi's teams, it catalogs the pharmaceutical group's websites. At the start, the team tracked 600. The tool ended up finding 17,000 domain names and 6,200 live websites, about 4,100 of which could move to post-quantum encryption. That figure comes from the case study Sanofi shared with LeMagIT, published on November 2: what a large group's preparation for the threat of future quantum computers looks like in practice.

The effort has been led since 2024 by Jérôme Lahalle, head of cloud and computing infrastructure cybersecurity. Work began after NIST, the US standards agency, published its quantum-resistant algorithms. A pilot followed in 2025, led by Alexandre Deloire, an expert in quantum and AI, with training for about a hundred members of the IT department. “Quantum risk has been built into the definition of risks at Sanofi,” says Jérôme Lahalle, who admits, however: “we know we are already late.”

Where to start when everything has to migrate?

Sanofi chose six criteria to decide the order of migration, in this order: legal obligations, application criticality, protocol maturity, availability of solutions from vendors, Internet exposure, and finally the type and strength of the encryption in place. Quantum risk appears in the risk map, and a dedicated clause has been added to supplier contract templates.

On the technical side, the team set up a test server and client to compare three types of certificates: fully post-quantum, hybrid (combining a classical algorithm and a post-quantum one) and backward-compatible. Size is one of the two issues identified, along with negotiation between the two endpoints. According to LeMagIT, moving from the classical ECDSA P-256 signature algorithm to the post-quantum ML-DSA-44 increases the signature from 64 to 2,420 bytes. Multiplied by the roughly 3 million certificates that Sanofi manages with EverTrust's platform, the effect is not trivial.

In 2026, the priority is SaaS applications, meaning software used online and hosted by its vendor. Sanofi assessed about twenty of them on a scale of 0 to 4: eight do not support post-quantum at all, five are at level 1, four have a solution that is ready and tested, with migration imminent, three have enabled post-quantum protections, and none is fully protected. Here Sanofi cannot do anything on its own: it is up to the vendor to migrate.

The same goes for network equipment. Zscaler's proxy does not yet support post-quantum, and it is expected by the end of the year. For VPNs, Sanofi is aiming for a hybrid key exchange and plans to update its Palo Alto Networks firewalls in the fourth quarter of 2026. Post-quantum is also part of the ongoing tender for its wide area network (SD-WAN).

What your organization can borrow

You don't need to be Sanofi to apply the method:

  • Inventory automatically rather than by hand: the gap between 600 known sites and 6,200 actual sites shows what a spreadsheet misses.
  • Ask your vendors and rate their answers on a simple scale, like Sanofi's.
  • Add a post-quantum clause to your contracts and tenders now.
  • Test certificate sizes on your equipment before switching over.

Our six-step guide to starting a post-quantum migration covers this approach in detail. One uncertainty remains, which Jérôme Lahalle sums up this way: “we don't know when quantum risk will become a reality.” For a group that depends on dozens of vendors, the timeline is only partly in its hands.

Method: AI-assisted monitoring, facts verified by a second AI model, article published automatically then reviewed and approved by Frédéric LENQUETTE on November 12, 2026. Our editorial policy.

Share this article