Mistral and DeepL executives warn against Europe's AI surrender

French artificial intelligence startup Mistral has released its latest model, Mistral Large 4, which reportedly surpasses Chinese competitors in coding capabilities. This development intensifies a growing dispute within Europe’s technology sector regarding the continent’s strategic direction. Leading figures argue against what they term a technological surrender, warning that relying solely on applications built on foreign models will erode Europe’s economic standing. Arthur Mensch, chief executive of Mistral, and Jarek Kutylowski, head of German translation firm DeepL, have publicly challenged policymakers who suggest Europe should focus exclusively on utilizing existing AI tools rather than developing frontier models. In a widely shared post titled "Build or surrender," Mensch warned that abandoning the creation of advanced models would cost Europe its remaining significance on the global stage. Kutylowski emphasized the necessity for Europe to master the entire AI value chain. He argued that failing to do so would cede economic benefits to other regions. Both executives cited the risk of new technological dependencies and potential trade deficits. They noted that software development costs are increasingly shifting toward AI models rather than engineering salaries. Without indigenous European models, this financial value would flow primarily to the United States. The debate contrasts with views held by some politicians, including European Commission President Ursula von der Leyen, who advocate for prioritizing clever applications over competing at the technological frontier. Max Junestrand, CEO of Swedish legal-tech company Legora, fueled the discussion by suggesting that building competitive frontier labs currently resembles wishful thinking. He recommended that Europe capitalize on available technology instead of chasing an uncatchable lead. Europe currently lags behind the US and China in developing complex AI systems, largely due to the immense computational power and financial resources required for training such systems. Despite this gap, the European Commission launched plans in summer to construct seven AI gigafactories valued at approximately 30 billion euros. Additionally, a consortium led by Italian firm Domyn is working on a proprietary frontier model intended to support all official EU languages. These initiatives aim to secure Europe's position in the evolving digital landscape.