Moving to AI-Native RAN

Why It Matters

AI‑native RAN can slash network operating costs and unlock new services, but only for operators that have already embraced cloud‑native, multi‑vendor Open RAN architectures.

Key Takeaways

• AI‑native RAN builds on existing Open RAN, not a full reset.
• Edge AI inference requires accelerators like Nvidia GPUs for real‑time tasks.
• Energy‑saving beamforming offers major OPEX reductions for operators.
• Multi‑vendor ecosystem needed to avoid fragmentation while fostering innovation.
• Cloud‑native infrastructure is prerequisite for AI RAN and future 6G.

Summary

The discussion centers on the transition from cloud‑native Open RAN to AI‑native RAN, with Wind River CTO Paul Miller explaining that the shift does not require a wholesale architectural overhaul but rather adds AI capabilities atop the existing Open RAN stack.

Key points include the need for edge‑focused AI inference engines—often accelerated by GPUs or specialized ASICs—to enable functions such as dynamic beamforming and power‑optimization. Miller stresses that AI introduces a "sense, think, act, optimize" loop, demanding continuous data collection, model updates, and real‑time latency guarantees that the underlying Kubernetes‑based, real‑time kernel infrastructure must meet.

He highlights that roughly 80% of RAN operating costs stem from antenna power, making AI‑driven energy‑saving techniques a prime revenue driver. Wind River’s Conductor and analytics suite already close the loop by gathering telemetry, refining models, and redeploying updates, illustrating a practical implementation of the AI lifecycle.

The broader implication is that operators must first adopt cloud‑native, multi‑vendor Open RAN platforms to unlock AI RAN benefits, while hyperscalers face challenges delivering low‑latency edge services. Successful AI RAN deployment promises substantial OPEX cuts, new B2B monetization avenues, and a foundation for the upcoming 6G era.

Original Description

The transition from cloud-native to truly AI-native RAN requires significant architectural shifts to harness the full potential of AI at the edge. Paul Miller, CTO of Wind River, discusses the missing architectural pieces, the path to deploying real-time AI models that meet strict service-level agreements (SLAs), and where the first scalable revenue opportunities for edge AI lie in the RAN domain. He also addresses how platform providers, such as Wind River, are evolving their offerings to enable AI-driven capabilities within the Open RAN ecosystem while mitigating fragmentation.

Featuring: Paul Miller, CTO, Wind River

Recorded May 2026

#telecomtv #thefutureofran #aiandml #openran #cloudnative #radioaccessnetwork #virtualisation #kubernetes #energymanagement #telcoandcsp #windriver #nvidia #dell #hpe #amd #intel