GERMANY MOVES TO SOVEREIGN AI
The German Ministry of Digital Affairs announced that the federal government has awarded a major contract to T-Systems, a subsidiary of Deutsche Telekom, and the software company SAP to develop a sovereign AI cloud platform. The contract involves an investment of 250 million euros, which will be split between the two companies. Deutsche Telekom will get 70% of the amount, and the rest will go to SAP.
“We are bringing a strategic decision to life: efficient digitization for the federal government, the Länder, and municipalities, built on infrastructure that we control ourselves—secure, scalable, and compatible with European standards,” argued Federal Minister for Digital Affairs, Karsten Wildberger. The German official added that the new platform will be “the backbone of a sovereign, digital administration with AI capabilities” that will help “speed up the country’s modernization.”
The German AI platform is designed as a central hub for the entire public administration sector, scalable, extensible, and interoperable. Once deployed, it will be a key component of the “Germany Stack,” a digital infrastructure of shared services among the federal government, the Länder, and municipalities.
DASSAULT SYSTEMS PARTNERS WITH PARISANTE CAMPUS, THE HUB FOR INNOVATION AND AI IN HEALTHCARE
Dassault SystEms provides its expertise in virtual twins and cloud technologies, while PariSanté Campus brings its ecosystem specialized in digital health innovation. Startups will have access to AI tools, a sovereign cloud infrastructure, and a mentoring program designed to turn innovative projects into solutions that meet European data protection and security requirements.
Beyond technological support, the partnership aims to address the main challenges faced by health startups: regulatory constraints, cost control, cybersecurity, data sovereignty, and the growing computational needs tied to artificial intelligence development.
For Antoine Tesnière, director of PariSanté Campus, this agreement should allow the campus’s players to have a sovereign and advanced data infrastructure, able to support the storage, analysis, and security of data under conditions suited to their international growth ambitions.
POST QUANTUM CRYPTOGRAPHY: TIME TO ACT
The French ANSSI (National Agency for the Security of Information Systems) has recently issued a formal statement that it will stop certifying products that don’t have quantum-resistant encryption starting in 2027. By 2030, only “quantum-safe” supplies will be required.Let’s be clear, these recommendations should be carefully implemented as ANSSI certification is mandatory for deployment within French administrations and for operating critical infrastructure. So-called ‘time capsule’ attacks, sometimes called ‘harvest now, decrypt later’ attacks, are already happening. Encrypted data captured on government networks, financial systems, or critical infrastructure can be stored and then decrypted once quantum capabilities have matured.This should not be seen as a simple supply issue. Encryption is built into networks, identity systems, software, databases, payment platforms, cloud services, and internal applications.
So, migration means mapping out cryptographic dependencies, assigning responsibilities, and coordinating changes between interconnected systems. This work takes time, no time to lose. The most effective starting point is to take an inventory of cryptographic assets: list where public-key algorithms are used, which systems handle long-term sensitive data, and which vendors can offer a credible migration roadmap.
QUANTUM COMPUTING: THE PARTNERSHIP BETWEEN PASQAL AND CREDIT AGRICOLE IS EXTENDED
Pasqal is a partner of Crédit Agricole and its subsidiary CIB since 2019. Together, the two companies are looking at putting the first use cases into production by 2028. To continue its experimental work started in 2021 with Pasqal, the bank has chosen the business topic of “monitoring capital consumption related to risk-weighted assets.”
CIB recalls having already conducted “several major projects” on other topics, including portfolio optimization. The bank emphasizes that it has thus demonstrated the ability of quantum algorithms to “outperform classical approaches,” but “under certain conditions.”
The results support “the relevance of a production rollout,” says CIB, and therefore the formalization of a roadmap toward production over the 2026-2028 period. This plan is structured around three main points. Goal: large-scale hybrid deployment. First, strengthening the capabilities of traditional infrastructure. An all-quantum approach is not on the agenda. CIB wants to deploy “quantum-inspired” algorithms on its existing IT infrastructure. The bank is thus counting on “immediate performance gains without using a quantum computer.” At the same time, it will run experiments on Pasqal’s neutral atom quantum machines. The goal of this phase is “to validate real performance under conditions close to a production environment.” Finally, the third focus is on large-scale hybrid deployment, combining quantum processing and classical high performance computing.
THE FINNISH IQM MANUFACTURER NOW LISTED ON THE US NASDAQ AS WELL AS ON THE HELSINKI STOCK EXCHANGE
On the week of June 29, the bell rang twice for IQM, once in New York as the company entered the NASDAQ, then in Helsinki.
Q-BIRD, A DUTCH QUANTUM DESIGNER, TO DESIGN QUANTUM PROOF COMMUNICATION
In the race to get ready for a world where quantum computers can break today’s encryption, the Dutch Q*Bird launches Falqon Key Manager, built to enable scalable, interoperable, hybrid quantum secure networks.
Ground State Ventures, an investor in Q-Bird, says : “Falqon combines Quantum Key Distribution (QKD) and Post-Quantum Cryptography (PQC) in one unified key management architecture. It delivers QKD keys, PQC keys, and hybrid keys across distributed network environments. It matters because a lot of what gets encrypted today, medical records, financial transactions, government communications, needs to stay secret for decades, not years. Data encrypted today can already be intercepted and stored by anyone patient enough to wait for the technology to catch up.The risk isn’t theoretical or future-only, it’s already happening to data being sent right now.”
We could not say it better: “harvest now, decrypt later” is not a theoretical risk. It’s time to start transitioning to quantum proof encryption and communication.
QUANDELA JOINS QUANTUM SOFTWARE ALLIANCE
QSA is a global alliance advancing quantum software and helping shape the path toward practical quantum computing. led by the University of Edinburgh, it champions and co-ordinates the development of quantum software and algorithms, to support quantum computing’s shift from lab-based theory to real-world applications.
As quantum hardware continues to progress, the software layer becomes essential to turn quantum processors into useful machines. QSA brings together leading research institutions and organizations working on key challenges such as quantum algorithms, compilation, verification, benchmarking, interoperability, and hardware/software co-design.
Says QUANDELA: “For us, this mission strongly resonates with our full-stack approach to photonic quantum computing. Our technology comes with its own native operations, noise models, compilation strategies, and application opportunities. Developing hardware, software, and use cases together is therefore central to building quantum systems that can deliver real value. By joining the Quantum Software Alliance, Quandela will contribute to a broader international effort to develop robust, open, and application-driven quantum software ecosystems — from algorithms and benchmarking to practical applications on emerging quantum platforms. We look forward to collaborating with the QSA community and contributing to the future of practical quantum computing!”
During the last QSA event, QUANDELA presented MerLin, a platform for discovering new quantum algorithms. As quantum capabilities advance, one key question is how to discover, evaluate, and scale algorithms that can fully leverage emerging quantum architectures.
MerLin is Quandela’s open-source discovery engine for photonic and hybrid quantum machine learning, designed to make quantum algorithm exploration more reproducible, benchmark-driven, and hardware-aware.
MerLin helps to:
Bring photonic quantum circuits into standard machine learning workflows
Test and compare quantum machine learning approaches in a reproducible way
Connect circuit design, model training, benchmarking, and future execution on cloud or quantum hardware.
The Post-Quantum Transition:
The European Moment
When is quantum computing arriving? I keep getting that question in dialogues with European executives this year, and it is the wrong one. Encrypted traffic is being collected today, to be
decrypted the day the machine becomes reality. Harvest now, decrypt later. I spent twelve years
running threat research across seventy countries, and persistent adversaries who collect first
and wait were the norm and not the exception. So the better question is actually around the
confidentiality horizon of your data. A contract, a research file, a financial exchange that must
stay confidential after 2030 is already exposed.
Why start now?
DORA has been in force since 17 January 2025, and the technical standard (Commission
Delegated Regulation (EU) 2024/1774) requires cryptographic capabilities that can progress as cryptanalysis advances. NIS2 expects essential and important entities to govern their cryptography deliberately. Only a few of them today could honestly claim to do so. The
European roadmap adds national roadmaps by the end of 2026 and transition targets in 2030 and 2035. The calendar is written, and these dates are here to stay.
A few things to pay attention to:
Post-quantum migration is not a one-time upgrade, and this is the part that boards need to understand and adapt to: it is a multi-year crypto-agility program.
Organizations that have attempted to build their crypto inventory describe it as more
difficult than the migration itself.
Then comes replacing the old stuff without breaking whatever depends on it, in addition
to proving both in audits. One blind spot deserves further attention: AI agents, which already exchange sensitive data over channels that were there before agents existed. Channel security comes inherited, from the CDN, from the gateway, from whichever provider terminates the connection. Inheriting can be
perfectly defensible, but only once someone has looked at it and confirmed the decision, with
an owner name next to it. As agentic systems spread through European businesses, the
quantum question and the AI question intersect exactly there, mostly unexamined.
Standards and vendors: the industrial stake. The post-quantum standards published by NIST are American by the authority publishing them,
though largely European by the science behind them. Eight of the ten designers of ML-KEM, the
primary encryption standard, worked at European institutions. HQC, selected in 2025 as a
backup algorithm, is largely of French origin. Europe did not import the technology, it is co-
designing it.
The economic risk is clear:
- Europe funds the research, then someone else is running what was invented. It
happened with cloud.
- The industrial value is not in the algorithms, as they are open, scrutinized and verified worldwide. The value is around them: trusted implementations, and above all the
inventory and migration work. Certification and the audit logs come after, and
regulators will ask for both.
- European providers can take that layer, but the buying decisions happen in the next two
or three budget cycles, not in 2030. Whoever is on the shortlist when a bank runs its first crypto inventory tender likely keeps the account for the decade.
What the ECA cybersecurity group is doing?
The ECA cybersecurity group started from that observation. Not a manifesto, a working method:
put European researchers, vendors and the organisations that buy from them in front of cooperation scenarios concrete enough to argue over, on AI as much as on quantum. The first
scenarios are already on the table. The hard part starts now, putting names and dates against
each of them. Europe has the science. The regulation is there, more of it than in any other market. What is missing is duller: agreed priorities, and something a CISO can point to when the board asks why the European option rather than the American incumbent.
Migration is happening either way; the regulators saw to that. Who does the work, and on
whose terms, is being decided now, in procurement cycles and in working groups like this one,
not in position papers.
Amin Hasbini is an AI & Cybersecurity Executive.
He works on post-quantum maturity and AI agent security. mahasbini.org
