Nokia made waves on July 15 with the launch of its AI-RAN platform, boldly claiming it as the industry’s first. This innovative platform, built on Nokia’s proven anyRAN software and powered by NVIDIA’s cutting-edge Aerial system, promises to revolutionize how operators extract capacity from their existing spectrum. It’s a move Nokia itself has described as one of the most significant shifts in radio architecture in decades, setting a new benchmark for network intelligence.
The technical ambition behind Nokia’s AI-RAN is impressive and directly addresses a critical industry need. Operators constantly seek to maximize their network efficiency without undertaking costly spectrum acquisitions. Nokia believes its AI-RAN platform provides a compelling answer, delivering substantial gains through intelligent software.
Unlocking Unprecedented Spectral Efficiency
Nokia isn’t just talking; they’re showing initial results and setting ambitious future targets. The company reports that its platform has already demonstrated over 20% spectral efficiency gains in early testing. Looking ahead, Nokia is targeting an impressive 50% improvement by 2027 and an astounding 100% increase by 2028, effectively doubling the capacity of current spectrum holdings.
These latter figures are significant targets rather than immediate results, reflecting a clear roadmap for future development. Nokia’s timeline indicates pilots will commence at the end of this year, with commercial availability anticipated in 2027. This phased approach allows for rigorous testing and refinement, ensuring the platform delivers on its grand promises.
One of the most appealing aspects for operators is the platform’s flexible deployment model, offered through a software subscription rather than requiring extensive hardware overhauls. This approach makes adopting AI-RAN more accessible and cost-effective. Operators will have three distinct options for integrating this powerful new capability into their networks:
- A GPU-powered plug-in card designed for seamless integration into existing AirScale sites.
- A standalone AI-RAN node, offering a dedicated, optimized solution for specific deployments.
- A cloud-server build, delivered through strategic partners, providing scalable, cloud-native functionality.
Nokia’s Strategic Pivot Towards AI
This launch is far more than just a new product; it represents a pivotal strategic shift for Nokia, particularly in its mobile radio business. Since taking the helm in 2025, CEO Justin Hotard has openly acknowledged that the mobile segment has struggled to deliver acceptable returns. Addressing this challenge has been one of his top priorities, leading to significant restructuring, including folding the mobile business into a new Mobile Infrastructure segment alongside broader cost-cutting initiatives.
Central to this revitalization effort is the groundbreaking partnership with NVIDIA, first announced in October 2025. This collaboration saw the chipmaker invest $1 billion for roughly a 3% stake in Nokia, cementing a deep strategic alignment. By leveraging NVIDIA’s cutting-edge silicon and CUDA software, Nokia can reduce its reliance on costly in-house custom chip development, redirecting valuable R&D resources towards software innovation – a transformation Hotard has emphasized as a move away from a legacy hardware-centric model.
Investors have responded positively to this strategic direction, with Nokia shares re-rating sharply throughout 2026, buoyed by the company’s growing momentum in AI and cloud technologies. The AI-RAN launch strategically landed just days before Nokia’s second-quarter results, further signaling its commitment to this new path. Industry analysts, like Omdia’s Rémy Pascal, are equally optimistic, projecting the cumulative AI-RAN market opportunity to exceed a staggering $200 billion by 2030. This indicates a clear and substantial market shift, and Nokia aims to lead the charge.
“Industry First” or Architectural Nuance?
Nokia’s claim of being the “industry’s first” AI-RAN platform warrants careful examination, especially when considering the competitive landscape. In June, just a month prior, Ericsson introduced its own commercial AI-in-RAN software subscription. Ericsson’s solution is already deployed across more than 15 live networks, demonstrating up to 20% higher downlink throughput and up to 10% better spectral efficiency.
Crucially, Ericsson’s offering runs on operators’ existing baseband silicon, requiring no additional GPU hardware, and is commercially available right now. Nokia’s claim to “first” therefore rests on a more specific definition: being the first GPU-accelerated AI-RAN platform. Both statements can be technically true simultaneously, highlighting a significant architectural divergence between the two industry giants.
This divergence runs deeper than mere timing. Nokia has inextricably linked its radio roadmap to NVIDIA, with its Chief Technology Officer, Pallavi Mahajan, acknowledging that some of the critical Layer 1 software is inherently bound to NVIDIA’s underlying hardware. Ericsson, by design, has taken the opposite route, ensuring its AI features remain silicon-independent to prevent such dependencies. While Nokia points to its broader ecosystem, including merchant silicon from Marvell, and highlights Open RAN compliance, the compelling performance gains it promises are currently rooted in NVIDIA’s proprietary stack, for which no direct alternative currently exists. This creates an intriguing tension between the messaging of “openness” and the engineering reality of NVIDIA dependency.
A Comeback in Motion
None of these nuances invalidate Nokia’s strategy. Outsourcing the intense and costly silicon race to NVIDIA, the dominant force in AI chips, offers a defensible solution to challenges Nokia previously faced in its radio business. Furthermore, embracing a software subscription model for radio infrastructure introduces recurring revenue streams, a significant improvement over the cyclical nature of traditional hardware sales.
However, it’s essential to remember that the platform is not yet commercially available, and its most impressive efficiency figures are still years away as targets. Moreover, a major competitor has already reached the market with a viable alternative via a different architectural path. For Nokia, this represents a strategic comeback in progress, not a victory already secured, and its trajectory is now undeniably and deeply intertwined with NVIDIA’s technology and market presence.
Source: AI News
