NVIDIA and Nokia have forged a strategic alliance to bring AI-RAN — Radio Access with Artificial Intelligence — from the lab to global commercialization, with a $1 billion investment by NVIDIA into Nokia and the launch of NVIDIA Aerial RAN Computer Pro (ARC-Pro), a “ready-for-6G” accelerated computing platform that unifies connectivity, computing, and sensing. Announced during GTC Washington, this partnership aims to establish 5G-Advanced and 6G native AI networks, with operators, hyperscalers, and manufacturers deploying racks and base stations capable of running AI algorithms alongside RAN functions.
The initiative is supported by tailwinds: industry estimates suggest that the AI-RAN market could exceed $200 billion by 2030, driven by the explosive surge of generative and agentic applications on mobile devices and at the edge. Meanwhile, T-Mobile US will collaborate with Nokia and NVIDIA to integrate and test AI-RAN technologies as part of its 6G innovation process, with field trials expected in 2026. Dell Technologies will supply PowerEdge servers as the computing backbone for Nokia’s AI-RAN solution.
What exactly has been announced
- NVIDIA–Nokia strategic alliance. Nokia will expand its RAN portfolio with commercial-grade AI-RAN products based on NVIDIA platforms.
- Investment: NVIDIA subscribes to new Nokia shares worth $1 billion (at $6.01/share), contingent upon typical closing conditions, representing approximately 2.9% of Nokia’s equity.
- New platform: NVIDIA ARC-Pro is introduced as a accelerated, “6G-ready” RAN computer; reference designs will be available for manufacturers and network vendors to create COTS or proprietary commercial products.
- Software: Nokia will accelerate deployment of its 5G and 6G RAN software on CUDA, extending its anyRAN approach (Cloud RAN and purpose-built RAN) and evolving AirScale with AI-RAN capabilities.
- Ecosystem: Beyond T-Mobile and Dell, the partnership extends to data networks with Nokia SR Linux on NVIDIA Spectrum-X and to telemetry and fabric management; both companies will explore Nokia’s optical technologies in upcoming NVIDIA infrastructure architectures for AI.
From 5G-Advanced to AI-powered 6G RAN
The transformative aspect of this alliance is moving AI from the cloud directly into the RAN itself. With ARC-Pro, base stations and edge sites can perform inference to:
- Optimize in real time spectral efficiency and energy consumption, link scheduling and control.
- Coexist user-side AI workloads (vision, XR, voice) with RAN functions, leveraging underutilized resources.
- Operate with software-driven innovation cycles, where new capabilities are delivered via updates, future-proofing towards 6G.
This decouples network evolution from hardware cycles, allowing 5G-Advanced to gradually evolve into native AI-enabled 6G without massive replacements. For operators, it means monetizing the edge — distributing inference across millions of nodes — and absorbing the growth of AI traffic with ultra-low latencies and data sovereignty.
Why now: AI traffic surges on mobile
AI usage on mobile devices is skyrocketing: services with hundreds of millions of weekly active users are nearly half used from smartphones, and app downloads of AI-powered applications exceed tens of millions per month. This pattern will intensify with native AI devices — XR glasses, drones, vehicles, robots — which demand edge computing and sensing with low latency. AI-RAN seeks to process data where it’s generated, reducing backhaul, costs, and carbon footprint per processed bit.
Roles of each participant
- NVIDIA: supplies GPU/DPUs, CUDA stack, frameworks (e.g., Sionna/AI Aerial for research and prototyping), and the ARC-Pro design as a 6G-ready baseline.
- Nokia: industrializes AI-RAN with its anyRAN portfolio and AirScale, ports RAN software to CUDA, and provides a software-defined path for coexisting new cards alongside deployed ones.
- T-Mobile US: plans to conduct field trials from 2026 to validate performance improvements and accelerate the shift toward 6G.
- Dell Technologies: provides PowerEdge servers as the computing platform for Nokia’s AI-RAN, featuring zero-touch upgrades and dynamic scaling.
Industry and geopolitical implications
The partners present an ambitious message: “AI-RAN” as a lever for the US to regain leadership in telecommunications, supporting manufacturing, integration, and deployment of 6G infrastructure with a U.S. stamp and a local ecosystem (operators, integrators, manufacturers). For Nokia, the agreement strengthens its data center exposure and its 6G ambitions; for NVIDIA, it extends its footprint from data centers to radio and edge as a computational fabric for the coming decade. The market responded with significant gains in Nokia’s stock following the announcement and details of the investment.
What it means for operators
- Efficiency and performance: AI algorithms enhance spectral utilization and reduce bit energy consumption.
- Deployment times: reference designs and certified racks speed up NPI and certification.
- Monetization: edge AI as a new business line (for inference services in industry, connected vehicles, retail, health), co-located with RAN on existing sites.
- Sovereignty: processing sensitive data locally helps comply with regulations and reduces data transfer.
Next steps and cautions
Field trials with T-Mobile are scheduled to start in 2026. The move to full deployment will depend on measurable KPIs (performance, energy, OPEX/CAPEX) and the maturity of Cloud RAN/purpose-built stacks with ARC-Pro. As with any industry-scale project, there are risks: multi-vendor hardware/software integration, thermal management in demanding sites, lifecycle automation of AI models, and the economics of mass updates to the installed base. Market forecasts and the actual closing of investments are subject to conditions and volatility.
Frequently Asked Questions
What is AI-RAN, and how does it differ from Open RAN or Cloud RAN?
AI-RAN integrates AI models into the RAN plane to optimize radio functions and host inference for services at the edge. Open RAN defines open interfaces among components, while Cloud RAN virtualizes functions in the cloud; AI-RAN can coexist with both, adding intelligence to the overall architecture.
What is NVIDIA ARC-Pro, and how does it fit with Nokia?
ARC-Pro is a 6G-ready accelerated platform combining connectivity, computing, and sensing. Nokia ports its RAN software to CUDA and integrates ARC-Pro into its AI-RAN solution, enabling operators to evolve from 5G-Advanced using software updates and compatible modules on AirScale.
What does NVIDIA’s $1 billion investment in Nokia mean?
This is a share subscription (about 2.9%) accompanying the tech partnership and aligning incentives for commercial development of AI-RAN. It is subject to usual closing conditions. The market responded with notable stock gains for Nokia.
When will users see “real” improvements in coverage or latency?
Initial pilot tests on commercial networks are planned for 2026 (T-Mobile US). The widespread rollout will depend on field results, software maturity, and deployment economics. The goal is to have AI-enabled 6G evolve from 5G-Advanced infrastructure via software updates and new cards.
Source: nvidianews.nvidia