ANALYSIS Two deals in quick succession — one ripping Palantir out of a French intelligence agency, the other routing hundreds of millions of euros through a Paris-based model maker into Saudi infrastructure — mark a concrete shift: sovereign AI is moving from policy language into procurement contracts, and European vendors are filling them.
Why it matters
ChapsVision, a little-known French software maker, won a contract in June to replace Palantir at France's domestic intelligence agency2. Separately, Mistral and HUMAIN announced a strategic collaboration spanning AI infrastructure, advanced model development, and the deployment of AI solutions in Saudi Arabia and across the region1,4,5. ◆ Together, these moves show that the "reduce reliance on US technology" impulse cited in the ChapsVision profile is translating into actual vendor substitution — not just in Europe but in Gulf states.
The big picture
ChapsVision is pursuing defence and intelligence contracts across the region as governments look to reduce reliance on US technology. The company targets an IPO by 2030, signaling confidence that the sovereign-mandate pipeline can sustain a public-market growth story. ◆ Winning the seat Palantir previously held inside a major Western intelligence service is the strongest possible proof point for that thesis: it demonstrates that a European vendor can meet the classification, integration, and performance bar that intelligence customers impose.
Mistral's Saudi deal operates on a different axis but serves the same strategic logic. The collaboration with HUMAIN represents a deal valued in the hundreds of millions of euros. Initial focus areas include cybersecurity and voice. The companies also plan to develop frontier models that perform strongly in Arabic language to help support the broader region. As part of the collaboration, Mistral will explore using HUMAIN's data center infrastructure to support growing local compute needs. HUMAIN and Mistral also plan to develop a joint go-to-market strategy in Saudi Arabia focused on bringing advanced AI solutions to regulated industries.
The collaboration is built so that data can remain within customer-defined boundaries, models can be adapted and owned on open weights, and training and inference can run on infrastructure and in jurisdictions the customer chooses. ANALYSIS That formulation — sovereignty as control over the learning loop, not merely data residency — defines a specific and demanding standard for what sovereign AI means in practice.
The two deals reveal complementary strategies for displacing US incumbents. ChapsVision is competing at the application layer — the analyst workbench and data-fusion platform that sits inside a classified environment. Mistral is competing at the model and infrastructure layer, co-developing frontier systems on local compute. One replaces an American platform already deployed; the other pre-empts American platforms from being deployed in the first place.
The financial structures differ in ways that matter. ChapsVision's path runs through government procurement and a future IPO, tying its fortunes to European defence budgets. Mistral's path runs through a Gulf sovereign-wealth ecosystem, with HUMAIN providing data-center infrastructure and a regional go-to-market channel. Both routes bypass US venture-to-enterprise pipelines entirely.
For Palantir specifically, the ChapsVision contract loss in France is notable. The stated rationale is strategic — reducing dependence on US technology — rather than performance-based.
What's next
ChapsVision's stated IPO target of 2030 gives the company roughly four years to convert sovereign mandates into recurring revenue at public-market scale.
For Mistral, the immediate deliverables are Arabic-language frontier models and cybersecurity applications built on HUMAIN infrastructure. The deal's value — hundreds of millions of euros — is large enough to materially shape Mistral's model-development roadmap while anchoring it to a non-US compute footprint. The collaboration was announced on August 24, 2026, from Riyadh.