Nokia
2026-07-29
Nokia and NVIDIA Launch First Commercial AI-RAN Platform with AI-Native Control Shift
Summary
Key Takeaways
Nokia introduced the industry's first commercial AI-RAN platform, core to which is the deep integration of its AI-native anyRAN software with the NVIDIA Aerial AI-RAN platform. The software-defined radio architecture supports existing Nokia AirScale radio units and Open RAN-compliant radios, with a clear upgrade path toward 6G. This architecture embeds AI inference directly into the RAN processing chain, aiming to improve spectral efficiency and reduce energy consumption.
On the management side, the MantaRay SMO platform gains Non-Real-Time RIC functionality and AI-enabled rApps for anomaly detection and dynamic network slicing. Agentic AI extends across IP, fixed, and optical domains. In hardware, Nokia launched AI-ready Doksuri Radios as an addition to the AirScale portfolio, with built-in AI acceleration. Nokia also partnered with Telia to develop AI-driven RAN use cases.
Why It Matters
Nokia's move is a strategy to defend against Ericsson and Huawei while encircling traditional RAN chip vendors like Intel by shifting compute from ASICs to NVIDIA GPUs. The deep integration of AI-native anyRAN with NVIDIA Aerial creates vendor lock-in to NVIDIA's GPU ecosystem and CUDA stack, undermining Open RAN openness.
Engineering limitations include tail latency from AI inference in real-time scheduling, and GPU power/thermal challenges at cell sites. MantaRay SMO's Non-Real-Time RIC cannot handle millisecond-level control, requiring complex coordination. The Doksuri Radios' AI acceleration capabilities remain vague, potentially overstating readiness.
PRO Decision
[Vendors: Ericsson, Huawei, Samsung] should accelerate AI-RAN development, emphasizing openness and portability to avoid NVIDIA lock-in. Support multi-vendor AI accelerators and open-source AI frameworks. Highlight advantages in real-time control and power efficiency.
[Enterprises: Operator CIOs and architects] must conduct zero-trust audits, assessing supply chain risks from AI-native anyRAN-NVIDIA Aerial integration. Demand AI model replaceability and data sovereignty guarantees. Independently test AI inference's impact on tail latency and base station power consumption, comparing TCO.
[Investors] should see through the hype: GPU deployment costs and power may offset spectral efficiency gains. The partnership may boost short-term stock, but long-term focus on operator adoption and competitor response. Beware of Nokia's capex pressure and NVIDIA's growing bargaining power.
Get 3-5 key AI infrastructure signals weekly →
💬 Comments (0)