Background: Media discourse on international trade is not neutral; it reflects ideological positions through language. Purpose: This study aims to analyze how Indonesian and Middle East media linguistically frame the 19% US tariff on Indonesian exports using Van Dijk’s Critical Discourse Analysis (CDA). Method: The research applied a qualitative descriptive design with Van Dijk’s CDA framework. Six online news articles published in July 2025 were purposively selected from major Indonesian (Tempo, Metro TV, The Jakarta Post) and Middle Eastern outlets (Al Jazeera, Sky News Arabia, Al Arabiya) based on relevance to the 19% tariff issue and narrative completeness. Articles were analyzed through Van Dijk’s macro–super–micro structure. Coded independently by two researchers using a shared coding guide before reconciling discrepancies through consensus, ensuring analytic reliability. Cross-checking linguistic patterns, thematic structures, and translations of Arabic texts further strengthened the credibility of the findings. Results and Discussion: The findings show that the six outlets cannot be classified by region, as their ideological orientations diverge within and across Indonesia and the Middle East. Indonesian media emphasize domestic stakes—technocratic caution, economic vulnerability, or sovereignty-driven critique—while Middle Eastern outlets embed the tariff within global power dynamics, ranging from anti-hegemonic (Al Jazeera) to pro-U.S. (Al Arabiya) or semi-neutral (Sky News Arabia). Across macro, super, and micro levels, ideological positioning arises from institutional agendas rather than geography. Shared strategies—evaluative lexis, power metaphors, and selective sourcing—function as socio-cognitive cues. That shape mental models of hierarchy and inequality, ultimately legitimizing specific interpretations of U.S.–Indonesia trade relations. Conclusions and Implications: The study shows that ideological framing of the tariff stems from outlet-specific discursive choices rather than regional identity. Language thus helps reproduce global power asymmetries. Future research should expand datasets and integrate socio-cognitive approaches to deepen analysis.