x
The global telecommunications industry is currently witnessing what could be one of the most transformative shifts in radio architecture in decades. On the 15th of July, Nokia unveiled its highly anticipated AI-RAN (Artificial Intelligence Radio Access Network) platform, built upon its anyRAN software and NVIDIA’s robust Aerial system. While Nokia boldly claims this to be an "industry first", the launch represents much more than a technological milestone; it is a critical strategic pivot for a historic brand fighting to reclaim its dominance in the mobile infrastructure sector.
At the core of Nokia’s technical pitch is a highly lucrative promise: allowing network operators to extract significantly more capacity from the spectrum they already own. The vendor reports that the AI-RAN platform has already demonstrated spectral efficiency gains of over 20 per cent. Looking ahead to its commercial rollout in 2027, Nokia is aggressively targeting a 50 per cent improvement, with ambitions to completely double the capacity of existing spectrum—a 100 per cent gain—by 2028.
Rather than forcing network operators into expensive, disruptive hardware replacement cycles, Nokia is offering this new capability via a flexible software subscription model. Telecommunications clients will be able to choose from three deployment paths: a GPU-powered plug-in card designed for current AirScale sites, a standalone AI-RAN node, or a cloud-server build facilitated through partner networks.
To view this merely as a product launch is to fundamentally misunderstand Nokia’s current corporate trajectory. Since taking the helm as chief executive in 2025, Justin Hotard has faced the monumental task of revitalising Nokia’s struggling radio division. Following a stark admission to investors late last year that the mobile business was yielding unacceptable returns, Hotard decisively folded it into a newly formed Mobile Infrastructure segment, accompanied by a sweeping programme of cost reductions.
The lifeblood of this turnaround is the alliance with NVIDIA, which was crystallised in October 2025 when the American chipmaker invested $1 billion for a roughly 3 per cent stake in Nokia. By leveraging NVIDIA’s merchant silicon and CUDA software architecture, Nokia has managed to drastically reduce its own costly internal research and development on custom chips. This pivot towards a software-centric model—and away from legacy hardware—has been warmly received by the financial markets, driving a sharp upward re-rating of Nokia’s shares throughout 2026.
However, the assertion that Nokia’s platform is the definitive "industry first" requires careful contextualisation. In June, Nokia's primary rival, Ericsson, successfully brought a commercial AI-in-RAN software subscription to the market. Ericsson's solution already boasts up to 20 per cent higher downlink throughput and a 10 per cent improvement in spectral efficiency across numerous live deployments. Crucially, Ericsson’s product operates on existing baseband silicon without the need for graphic processing units (GPUs).
Nokia’s claim to being first hinges on a much narrower, specific definition: it is the first GPU-accelerated AI-RAN platform. This fundamental difference highlights a sharp divergence in corporate philosophy. While Ericsson has intentionally kept its AI features independent of specific silicon to avoid hardware lock-in, Nokia has tethered its radio roadmap tightly to NVIDIA. Nokia’s chief technology officer, Pallavi Mahajan, has openly acknowledged that a portion of their Layer 1 software is inextricably linked to the underlying NVIDIA hardware. Although Nokia promotes the platform as Open RAN-compliant, the profound efficiency gains it is selling currently rely entirely on NVIDIA’s stack—an ecosystem for which no direct alternative exists today.
Outsourcing the silicon development race to the world's undisputed leader in AI chips is a highly pragmatic, defensible move for a business unit that had previously struggled to innovate in isolation. Furthermore, transitioning to a subscription-based software model provides the mobile division with the recurring revenue streams that traditional hardware lifecycles simply could not sustain.
Yet, as industry analysts at Omdia project the cumulative AI-RAN opportunity to exceed $200 billion by 2030, Nokia’s true test still lies ahead. The platform is not yet fully commercialised, its headline efficiency numbers are still two years away from becoming a reality, and formidable rivals have already reached the market via alternative, hardware-agnostic routes. Nokia’s AI-RAN launch is not a victory lap, but rather a compelling comeback in motion—one whose ultimate success is now permanently tethered to NVIDIA.
For readers looking to explore the original reporting and gain more detailed insights, visit Artificial Intelligence News:
👉 Nokia’s AI-RAN platform: a radio comeback that runs on NVIDIA
Disclaimer: This article is provided for informational purposes only, mistakes may be made, and it's not offered or intended to be used as legal, tax, investment, financial, or any other advice.
