AI-native by design: Building the foundations for 6G

As the telecom industry accelerates toward 6G, the role of AI is undergoing a fundamental shift. No longer just an enhancement layer, AI is becoming foundational to how next-generation networks will be designed, deployed, and operated. Leaders from Deutsche Telekom, Aira Technologies, and Rakuten Symphony recently explored the architectural, operational, and cultural shifts required to build AI-native 6G infrastructures – where intelligence is embedded into every layer of the network.

Speakers:

• Mohammad Sharifan – Lead Architect, Group Technology / Data & AI, Deutsche Telekom
• Geoff Hollingworth – CMO, Rakuten Symphony
• Rahul Atri – Chief Product Officer, Aira Technologies

Watch the full interview.

The panel opened with a shared perspective: traditional network architectures and rules-based applications are not equipped to meet the complexity of 6G. AI-native networks demand a re-architecture – one that enables distributed inference, real-time data pipelines, and intelligent lifecycle management. These are not just technology upgrades, but structural changes in how networks function.

Programmable infrastructure, autonomous decision-making, and AI-driven agents will replace rigid workflows. But this shift also calls for a new operational mindset – where organizations themselves become AI-native, designed around human-agent collaboration and intelligent automation from the ground up.

Operational complexity, organizational inertia, and the need for trust

While the technology foundations for AI-native are maturing, telcos must also contend with organizational fragmentation and legacy silos. The panel emphasized that without unified data models and shared ontologies across vendors, AI adoption will remain patchy and hard to scale.

More than just deploying advanced models, telcos must also address issues of trust and alignment – ensuring AI-generated decisions reflect business intent, deliver measurable value, and are auditable. This is especially critical as human oversight gives way to agent-led operations.

What emerged was a strong call for clarity, not just in technical design, but in how success is measured. Without consistent targets or industry-wide benchmarks, even the best AI implementations risk becoming siloed experiments rather than scalable solutions.

Key takeaways

• AI-native is not an upgrade – it's a redesign. Telcos must rethink network architecture, data flows, and operations to support embedded intelligence.
  
• Distributed intelligence needs clean, harmonized data. Without common data models, scalable AI is out of reach.
  
• Organizational silos are the real roadblock. Technical capability exists – cultural and structural change is the bigger challenge.
  
• Trust is built on outcomes. Measurability, explainability, and governance are essential to scale AI use in operations.
• Forget the “Gs.” The shift to AI-native is bigger than any generation – it's about making networks programmable, autonomous, and value-driven.
  
  

"So I think this workforce that's half human, half machine, is the future. But we're going to have the same problem with AI agents as we have with human agents."

-Geoff Hollingworth, CMO, Rakuten Symphony

Follow us for more insightful perspectives on the future of cloud-native, AI-native, and programmable telecom networks.