Nokia Set to Reshape Telecom Infrastructure with GPU-Powered AI-RAN Advances

Nokia has launched its breakthrough commercial AI-RAN platform, combining its AI-native anyRAN software with NVIDIA’s Aerial computing architecture to modernize mobile network infrastructure. This collaboration transitions radio access networks into programmable AI platforms capable of doubling spectrum capacity and boosting spectral efficiency by over 100% by 2028. By enabling deployment on existing hardware alongside ORAN-compliant radio units, the system unlocks higher throughput without forcing operators into costly full-hardware replacements.

The architecture introduces a software subscription model, enabling carriers to receive continuous feature updates and performance improvements at software speed. Through this approach, we anticipate that telecommunication providers can dramatically lower cost-per-bit operations while establishing an evolutionary bridge toward 6G capabilities. The joint initiative positions the radio access network as critical computing infrastructure, transforming passive cellular connectivity into an active, planet-scale AI compute engine.

We define AI-RAN (Artificial Intelligence Radio Access Network) as a software-driven architecture that embeds high-performance GPU computing directly into baseband stations to run radio functions, perform dynamic signal processing, and process edge AI workloads on the same physical infrastructure. By combining Nokia’s anyRAN software with NVIDIA’s accelerated computing platforms, it transforms passive cell towers into an active compute engine that optimizes radio performance dynamically in real time. While we have interacted with some mobile operators raising concerns regarding power consumption and the energy footprint of hosting energy-intensive GPUs at edge sites, we see Nokia countering and resolving these concerns through software-driven efficiency gains.

Nokia’s platform runs compute-intensive AI algorithms, such as deep receivers and non-linear channel estimation, that achieve over 100% spectral efficiency improvements, operating within existing power envelopes while drastically lowering the overall energy cost per bit transmitted. Additionally, Nokia incorporates AI-powered energy management algorithms that dynamically scale compute resources and power down underutilized components based on real-time traffic demand. By pairing these predictive power-saving routines with the ability to monetization-share compute capacity across both telecommunication and edge AI inference workloads, Nokia shifts the equation so that the operational and energy output yields a significantly net-positive return for network operators. .

Nokia’s Software-Defined AI-RAN Architecture: Open RAN, NVIDIA GPU, and Ecosystem Integration

Nokia's AI-RAN platform addresses diverse operator strategies by providing a single software-defined foundation powered by anyRAN software and NVIDIA accelerated computing across three distinct adoption pathways. By supporting 4G, 5G, and open multi-vendor Open RAN standards, the architecture enables operators to modernize their infrastructure incrementally without risking vendor lock-in or prematurely stranding deployed assets. Existing Nokia clients can integrate advanced AI capabilities through a plug-in GPU expansion unit, boosting AirScale baseband performance while extending the overall return on past hardware investments.

Today the mobile ecosystem incorporates AI-accelerated merchant silicon from partners such as Marvell to deliver a flexible software-defined network that continuously evolves through rapid software updates. By decoupling architectural modernization from forced hardware replacement cycles, Nokia enables operators to scale cloud and edge capacity at their own strategic pace. This multi-path deployment model lowers adoption friction, establishing an evolutionary bridge toward next-generation network performance.

From our viewpoint, Nokia can also deftly address other objections to AI-RAN, specifically around capital expenditure, infrastructure lock-in, and operational disruption, by leveraging its flexible three-path architecture, multi-vendor Open RAN compliance, and software-defined commercial model. Rather than obliging a costly rip-and-replace upgrade cycle, Nokia enables operators to introduce GPU-accelerated computing incrementally through plug-in cards for existing basebands, standalone nodes, or commercial cloud-native servers, preserving existing network assets while scaling compute at their own strategic pace. By decoupling software features from physical hardware updates and fully supporting Open RAN standards, Nokia alleviates concerns over proprietary vendor lock-in, enabling operators to continuously evolve their networks to 6G-ready performance purely through software subscriptions.

Nokia’s GPU-powered standalone nodes and COTS server solutions eliminate architectural rigidities by giving operators the freedom to scale AI-native capacity across proprietary basebands, clustered configurations, or industry-standard cloud environments. By integrating 4G, 5G, and future 6G workloads into a single logical base station framework, the platform protects infrastructure investments while facilitating cross-generational network upgrades. We find that partnering with ecosystem vendors for commercial off-the-shelf hardware ensures a secure, open supply chain that combines cloud-native agility with high-performance radio processing.

We see Nokia’s subscription-based AI-RAN model shifting mobile networking from periodic, capital-intensive hardware refresh cycles to continuous software-driven performance updates, significantly reducing total cost of ownership. By delivering real-time spectral efficiency improvements, dynamic optimization, and 6G-ready capabilities through software activations, the platform transforms traditional network infrastructure into a scalable, long-term asset that maximizes return on investment without hardware premiums.

Outpacing Legacy Silicon: Nokia’s Strategic Edge in the Commercial AI-RAN Race

In the emerging AI-RAN market, Nokia's primary competitors include Ericsson, Huawei, and Samsung Electronics, alongside cloud-first RAN ecosystem players such as Fujitsu. While traditional rival Ericsson initially focused on proprietary custom silicon (ASICs) or software-only automation layers on existing hardware, we identify Nokia gaining a significant competitive edge through its early, deep strategic alignment with NVIDIA to bring accelerated GPU computing into commercial baseband processing.

By using NVIDIA’s Aerial platform alongside merchant silicon partners such as Marvell, Nokia breaks away from costly vendor-locked ASIC cycles while outperforming competitors on raw computational throughput for complex AI deep-receiver algorithms. By coupling a $1 billion financial backing with direct integration into NVIDIA’s CUDA ecosystem and Aerial platform, NVIDIA's strategic investment allows Nokia to bypass the costly, vendor-locked ASIC development cycles that constrain competitors like Ericsson, providing an immediate performance and economic advantage through GPU-accelerated, software-upgradable 6G readiness

From our viewpoint, Nokia establishes a structural advantage over legacy vendor architectures through its multi-path deployment options, offering GPU-powered plug-in cards, standalone AI-RAN nodes, and cloud-native COTS server options that fully adhere to multi-vendor Open RAN standards. As a result, Nokia transforms telecom capital expenditure by pioneering a value-based software subscription commercial model, enabling operators to activate over 100% spectral efficiency gains and 6G-ready capabilities on-demand without paying full hardware premiums.

Specifically, we find that Nokia establishes a competitive advantage over Ericsson by deploying accelerated GPU computing directly into baseband processing, whereas Ericsson relies on CPU- and ASIC-based software enhancements. While Ericsson's AI in RAN software achieves incremental spectral efficiency gains of around 10% to 15% on existing hardware, Nokia's NVIDIA-powered anyRAN platform targets over 100% spectral efficiency improvements by 2028, doubling capacity on existing operator spectrum. By marrying high-performance GPU hardware with flexible multi-path deployment options, Nokia provides a more discernible scalable, computationally advantageous path to handle complex 6G algorithms and enterprise edge AI workloads.

Further underpinning its sales and marketing momentum, Nokia’s AI-RAN commercial timeline targets pilot trials by year-end 2026 and market availability in 2027, anchored by T-Mobile US as the lead deployment partner. Alongside this lead relationship, Nokia maintains an active global ecosystem of engaged operators, including Deutsche Telekom, Vodafone, Orange, BT, Elisa, SoftBank, NTT DOCOMO, and Indosat Ooredoo Hutchison.