Orange exec outlines the three ingredients for production-grade AI

• Orange’s Philippe Ensarguet outlined what operators need to scale AI deployments
• Speaking at the DSP Leaders World Forum in Windsor, he set out his three key AI ingredients
• Cultural shift is just as important as tech upgrades for operators

WINDSOR, UK – DSP Leaders World Forum 2026 – Telecom operators looking to scale AI successfully must focus on organisational readiness, operational discipline and production-grade execution – not just the underlying technology, Orange VP of cloud and software engineering Philippe Ensarguet told industry leaders at TelecomTV’s annual DSP Leaders World Forum event.

During a keynote address as part of a panel entitled ‘Scaling AI Deployments’, Ensarguet said operators need the right “organisational wiring” to support AI at scale, highlighting the importance of operating models and cross-functional teams that combine business expertise with deployment and operational capabilities.

Ensarguet outlined what he sees as the three essential ingredients for production-grade AI in telecom: Clean data; programmable infrastructure; and AI outputs connected to real operational decisions and accountable business owners.

“Most of the time, when I look at what is going on in the industry, at least one of the three is missing,” he said, warning that this is preventing operators from moving AI initiatives beyond experimentation and into scalable operational deployment.

The Orange executive said the industry conversation is now shifting from AI exploration to practical execution. He noted there is a need for clearer architectural patterns, stronger operational models and better communication of AI’s business value to executives and ecosystem partners.

He also warned that telecom operators remain too heavily driven by “tech push”.

“Most of the time, people are not selling solutions, they are selling technology,” he said. “We need to focus first on the value we generate and the business case before starting.”

With Orange running operations in more than 26 countries across Europe, the Middle East and Africa, Ensarguet was well placed to share his experience of the challenges faced by multinational telecom groups looking to scale AI across their entire portfolio. He argued that operators need a standardised “AI blueprint” built around common architectural patterns and non-negotiable building blocks.

On the cultural side, he said many organisations still have unrealistic expectations about AI capabilities and need better education in terms of where machine learning, traditional AI and generative AI can deliver practical value.

“We need to train people and get their hands dirty with AI,” he said.

He was joined on the panel by: Ahmed Hafez, SVP of network strategy and data & AI in networks at Deutsche Telekom; Amol Phadke, chief transformation officer at Tech Mahindra; and Ohad Barnoy, head of GenAI and data solutions, services and consulting at Amdocs. Watch out for further coverage of this panel discussion and the on-demand recording of the discussion, which will be made available, along with all the other sessions from DSP Leaders World Forum, on TelecomTV in the coming days.

- James Pearce, Editor, TelecomTV