Opinion ·
Your AI is expensive, and your CFO wants to know what it delivers
Original title, in French: “Votre IA coûte cher, votre DAF veut savoir ce qu'elle rapporte”
- Publication
- Journal du Net (France)
- Author
- Guillaume Brandenburg
- Published on
- 21 September 2026 · read in the page structured data
Summary
The article describes a strategic shift for French companies on AI in autumn 2026: after a phase of large-scale, disorderly experimentation, the aim is to move from "tokenmaxxing" (a race for usage volume) to "valuemaxxing" (measuring actual return on investment). According to an Accenture study cited in the article, companies that have prepared their proprietary data for AI report an operating profit 4.5% higher than their peers. The author identifies three conditions for making these investments pay off: starting from business processes rather than models, ensuring data quality and governance, and favouring frugality by using lightweight models for repetitive tasks rather than systematically using the most powerful ones.
Summary based on the full text of the article.
Read the article on Journal du Net →Figures cited
-
4.5 %
higher operating profit for companies that have prepared their proprietary data for AI
companies that have prepared their proprietary data for AI, compared with their peers
« Selon une étude d'Accenture, les entreprises qui ont préparé leurs données propriétaires pour l'IA affichent un résultat d'exploitation supérieur de 4,5 % à celui de leurs pairs. »
Source: Accenture
Classification
- Country
- France
- Company size
- All sizes
- Industries
- All industries, or industry not specified
- Companies
- Accenture (study author)
- Keywords
- return on investment, data governance, financial management, AI adoption, CIO, executive committee
Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to Journal du Net and its author. Entry translated from French by Claude Sonnet 5.5.