The world of artificial intelligence is buzzing with the arrival of DeepSeek R1, a revolutionary open-source model developed in China. This bold project challenges the dominance of established leaders like OpenAI, thanks to remarkable technical performance and unprecedented accessibility. Indeed, unlike proprietary models, DeepSeek R1 relies on the complete openness of its code and reduces usage costs to a fraction of those offered by its competitors. As the boundary between closed innovation and open-source solutions blurs, this advancement raises many questions: does it mark a sustainable democratization of AI or a disruption of the economic balances in the sector?