Arthur Mensch warns dependence on US systems could be abruptly severed.

Arthur Mensch, chief executive of French AI developer Mistral AI, urged European governments to steer defence and security spending towards homegrown artificial intelligence at the Tallinn Digital Summit in Estonia on Thursday. He argued that Europe must control core technologies to preserve geopolitical clout and reduce exposure to decisions made in other jurisdictions.

He cautioned that reliance on foreign platforms leaves militaries vulnerable to cut-offs at critical moments, citing lessons from the war in Ukraine. Reports that Elon Musk curtailed Starlink satellite coverage during a Ukrainian operation have intensified European concerns about the resilience of outsourced digital infrastructure, while political wrangling in Washington has raised doubts over the long-term reliability of US support for Kyiv.

Mistral AI, a Paris-based builder of foundational AI models, has emerged as one of Europe’s leading challengers to US and Chinese AI groups. In February, it formed a strategic partnership with Helsing, a European defence technology firm, to integrate large-scale AI models with sensor systems for drones and other battlefield uses.

Mensch’s intervention comes as American technology companies broaden their overtures to European defence customers. Meta has opened its Llama AI model to European governments for security applications, highlighting the depth of US offerings. He countered that public funds for defence AI should prioritise European-headquartered suppliers to ensure continuity of access and stronger legal and political accountability.

The push for sovereign capability reflects a wider debate in European capitals over how to balance openness to global innovation with strategic autonomy. For advocates of a homegrown approach, the conflict in Ukraine has turned that discussion into an immediate procurement challenge.