Issue #21 - May 29, 2026
Mistral's message to lawmakers: no EU tech sovereignty without Mistral
Sebastien Louradour
5/29/20264 min read
Mistral's message to lawmakers: no EU tech sovereignty without Mistral
Mistral's assertiveness towards its US competitors has become much stronger recently and played out in Arthur Mensch's testimony before the French National Assembly's inquiry commission on digital vulnerabilities, where Mistral's CEO assured that France was about to become a vassal state. His position was again confirmed two weeks later, when Mensch showed little enthusiasm for the Pope's latest encyclical on AI: "we're all for peace but (...) as long as we have adversaries that are threatening, and they are threatening, we need to have our own capabilities. Read alongside Mistral's April 2026 white paper, the hearing shows the stakes the main European AI Lab is facing, how it presents it to a political audience, and what it could mean for the ongoing AI policy debates in Europe, in particular files such as the EU Cloud and AI Development Act (CAIDA).
Mensch's approach fully embraces the ongoing push towards technology sovereignty of Europe, and suggests a policy agenda that would position Mistral at the center of a European AI plan. He lists the potential threats Europe is facing from US incumbents such as a de facto dominance in future data center investments and usage, and ultimately a risk of seeing the value created by tokenization flow back to the US instead of the EU. According to him, Europe is facing a "colonisation" threat from foreign technology, with the risk of seeing Europe merely counting as a big market instead of an industrial powerhouse.
Ironically, while we could assume that such a view should easily convince lawmakers that Mistral is the French champion to save, some reactions over the audience and afterwards have been quite chilling, particularly regarding the main message of Mensch which essentially asks to unleash massive and unprecedented AI investments. In a video, the rapporteur of the Parliamentary Commission, Cyrielle Chatelain, a MP from the Green Party, has accused Mensch of staging a gap between those who are benefiting from the tech revolution and those who are not. Philippe Latombe, from the centre-right party MoDem, and one of the rare MPs showing up at the hearing questioned the CEO over environmental consequences of data center building and energy consumption, showing little sympathy for the French champion. This confirms that the tech backlash, and the scepticism towards innovation hasn't faded among some French policymakers, who mostly see technology as a social divider and a threat rather than an opportunity, even if the player can bring in tax revenue and domestic growth. But Arthur Mensch had other messages to share to lawmakers, in particular on sovereignty where France's posture has become increasingly assertive — at times prioritising the sovereignty framing over the harder question of what industrial strategy would actually sustain it. While the CEO acknowledges that sovereignty can't be achieved without initial contributions from outside the EU and welcomes for instance foreign capital investments as a necessary way to circumvent the lack of capital access in Europe, most of his focus though is rather about how to build sovereignty out of industrial capacity to boost capacity for Europe to invest in R&D.
When it comes to compute capacity, his diagnostic is staggering: without a significant push to provide more energy, Europe is likely going to serve the AI Labs that move the fastest, pointing out to a risk of transfer of value outside of Europe. Mensch's words to summarise the game are utterly simple: the highest margin lies in the capacity to transform energy into tokens, and requires unprecedented investments in energy grids to keep up with the growing demand of users. Mensch also describes the current limits of the EU digital single market: navigating different legal entities, stock option regimes, labour regulations, and banking requirements slows down the capacity to scale across Europe. He also considers that this status quo ultimately prevents large investments and R&D from the private sector. Mensch illustrates this with a striking comparison — Europe has approximately 60 telecoms operators against 3 in the United States. The consequence is a fundamental capital disadvantage. US telcos can commit vastly larger AI investment budgets precisely because their margins are not diluted across fragmented national markets. When a US startup sells to three large telcos, it reaches scale rapidly, recycles venture capital efficiently, and then exports to Europe — where by that point no comparable domestic champion has emerged.
Translated into EU regulation, Mistral advocates for the future Cloud and AI Development Act (CAIDA) to embed specific European AI criteria in public procurement for compute infrastructure, including open and interoperable AI models, data residency guarantees, EU jurisdictional control, energy efficiency standards, and AI-readiness requirements. Crucially, it proposes a European training requirement: above a defined compute threshold, AI model training should be performed exclusively on EU-certified infrastructure, initially in the public sector and high-criticality use cases. This incentive model involves public procurement but not only: Mistral calls for a 20% tax credit for businesses adopting certified European AI infrastructure (30% for SMEs and strategic sectors), and compute vouchers distributed through national promotional banks such as Bpifrance, KfW, and CDP.
Mistral could indeed benefit from a buy European approach on AI Models and Cloud infrastructure. The open source AI model approach Mistral has built since its start represents a strategic card. By calling for open source solutions for their competition and interoperability benefits, the EU can indirectly favour Mistral without having to call out the main US heavyweights that develop primarily close models. The Tech sovereignty and open source packages of the EU should be relieved next week after having been postponed a few times already. In a leaked version consulted by journalists, the draft already confirms a significant move towards a "de-risking" approach towards foreign technology. The Commission for instance suggests securing supply chains for semiconductors in critical sectors such as weapons and medical devices to avoid shortages. According to the FT, in the draft document, Brussels acknowledges that the bloc is "almost entirely dependent on the US and Asia" for the most advanced chips. In a modern world made out of globalised supply-chains and inter-dependencies, the journey to reach sovereignty may well prove unaffordable.