Orange - A Case Study in Telco Automation, AI and AN: Managing Sovereign AI and Network Automation for 300 Million Subscribers across 26 Nations from Senegal to Spain

This profile addresses the role of automation, autonomous networks (AN), and AI/GenAI in Orange's telco operations. The goal of this series is to answer how deployment of these technologies can cut costs and create new revenues, and identify which telcos are achieving tangible margin gains from their investments.

Orange is France's incumbent and largest telco, operating across Europe and MEA. It ranks 11th globally by revenue, top three in Europe, and top five in MEA. 2025 revenues were $45.6B, up 2% YoY, though long-term growth is flat relative to 2015 revenues of $44.7B. Annualized EBIT margin declined from 16.2% in 4Q24 to 13.2% in 3Q25, below the 16.5% global average. Labor costs exceed 30% of opex (ex-D&A), compared to the low 20% range for the global market. Despite reducing headcount 9% between 3Q22 and 3Q25 to 124K employees, EBIT per employee was $46.8K in 3Q25, up from $44.5K in 3Q22. During this period, the global average EBIT per employee increased 16% to $66.9K. For Orange, high labor costs and recently declining margins suggest that headcount reductions are not enough to close the productivity gap. This cost pressure boosts the case for investment in automation and AI.

Orange has built real automation infrastructure. Its Pikeo 5G SA experimental network and the Network Integration Factory that followed it are deployed and producing real results, not just slideware. The Euro-300 million in AI-generated value reported for 2025 is tracked against a group-wide dashboard with country-level accountability. The harder question is whether these gains can scale. Orange operates across 26 countries at very different levels of network maturity. Its own CTO acknowledged in 2023 that most affiliates were starting from approximately TM Forum automation Level 2. Closing that gap by 2028 while simultaneously decommissioning copper in France, managing a restructuring in Orange Business, and building out AI revenue streams is an ambitious program. The primary risk is not that the technology fails but that the complexity and cost of running multi-vendor, AI-driven networks in heterogeneous markets offsets the efficiency gains that justify the investment.

Organizations mentioned

• Amdocs
• ARCEP
• Augtera
• AWS
• Casa Systems
• Ciena
• CSG
• Dell Technologies
• EE
• Ericsson
• Google Cloud
• HPE
• Jazztel
• LangChain
• MasOrange
• Mavenir
• Meta
• Mistral
• Nokia
• Nvidia
• OpenAI
• Samsung
• SUSE/Rancher
• Totem Towers
• Voo