The Future of RAN

Orange warms to Nokia’s AI-RAN pitch

By Ray Le Maistre

Apr 15, 2026

  • With Nvidia as an investor, Nokia is pushing AI-RAN architectures even harder
  • The vendor is gaining traction as it claims efficiency advances with its combination of its RAN software with Nvidia GPUs
  • Orange will work with Nokia and Nvidia to explore how such AI-RAN systems could improve performance and support emerging services, such as sensing

Orange is the latest major operator to explore the full potential of AI-RAN, including the deployment of GPU (graphics processing units) in the radio access network, with Nokia and Nvidia

The AI-RAN concept has been high on the mobile network operator (MNO) agenda for more than two years, following the launch of the AI-RAN Alliance in February 2024: That industry body had a number of big name founding members, including Nokia and Nvidia, but didn’t attract a lot of initial support from the MNO operator community because one of its key propositions was the deployment of GPUs in the RAN, a concept that rang a lot of cost and energy consumption alarm bells with MNO executives. 

It’s key to note that the alliance has three main propositions: 

  • AI-for-RAN – the use of AI tools to improve the performance and efficiency of radio access networks. This is something that all mobile operators are exploring and is not contentious.  
  • AI-on-RAN – the development and deployment of AI-enabled applications at the edge of the network that can be delivered over 5G connections. This is a concept that is still largely unexplored as it is closely linked to an upgrade in compute and storage capabilities in the RAN.  
  • AI-and-RAN – the integration of AI and RAN processes on the same underlying infrastructure, so that the resources are constantly being used even if there is no mobile customer activity on the network, though with RAN processes prioritised when there is network activity. This has been the main area of contention, as the primary approach to enable ‘AI-and-RAN’ has been linked to the deployment of GPUs that, many argued, would be too expensive and demand too much energy consumption.

But that was two years ago and a lot has changed since then. 

Now, more operators are exploring the potential of AI-RAN in general and AI-and-RAN in particular, most notably with Nokia, following Nvidia’s investment in the Finnish vendor in late October last year and the subsequent focus by the partners on the development of an AI-and-RAN solution that is more engineered towards the needs of, and therefore more palatable to, MNOs – see Nvidia lights a fire under AI-RAN partner Nokia with $1bn investment.

That AI-and-RAN solution is the combination of Nokia’s anyRAN software and the Nvidia Aerial RAN Computer Pro (Arc-Pro), which the AI technology giant describes as “a 6G-ready telecommunications computing platform… that combines connectivity, computing and sensing capabilities”. 

During the recent MWC26 show in Barcelona, Nokia focused heavily on AI-RAN advances and its related engagement with an increasing number of operators, including BT, Elisa, NTT Docomo and Vodafone Group. (You can learn more about how Elisa is approaching AI-RAN in a session from our recent Future of RAN summit – see RAN architecture for the AI-native 6G era.

Now Orange, a long-time customer of Nokia, with which it has also been exploring cloud RAN options (see this update from February 2025), has agreed to work with Nokia “through a structured co‑innovation framework” to “jointly co‑create and co‑innovate AI‑RAN use cases, supported by Nvidia AI infrastructure” and the telecom vendor’s anyRAN 5G software. 

Specifically, “Nokia and Orange will jointly identify, design and evaluate new AI‑RAN capabilities,” noted the companies in this announcement. “The collaboration aims to explore how a GPU-based radio processor can boost radio performance with more advanced receivers, and how AI can be tightly integrated into the RAN to further improve performance, support new services, such as sensing [integrated sensing and communication, or ISAC], and bring greater automation and intelligence to both cloud-based and purpose‑built RAN environments.” In AI-RAN Alliance parlance, that would be a combination of AI-for-RAN and AI-and-RAN.

And Orange is also looking ahead to the next generation of mobile network infrastructure. “As mobile networks evolve towards 6G, Nokia and Orange will co‑develop approaches to maximise the spectral efficiency of existing and future bands, including the upper 6 GHz band,” which has been identified by the European Union’s Radio Spectrum Policy Group (RSPG) as the primary band suitable for the introduction of 6G in Europe by 2030, as TelecomTV reported earlier this year. “The 6G-ready platform will enable a smooth, software-defined migration to 6G and support smarter use of compute resources across Orange’s operational footprint,” the partners noted. 

They added: “By working with Nokia and Nvidia, Orange aims to deepen its understanding of how AI-enabled RAN functions can be integrated seamlessly into operational networks while ensuring sustainability and efficient resource utilisation across Europe, the Middle East and Africa.”

Orange’s group CTO, Laurent Leboucher, noted that the operator “is committed to building more efficient, adaptable and sustainable networks. By collaborating with Nokia and Nvidia on AI‑RAN, we can better understand how the AI-native architecture enabled by AI-RAN can improve the efficiency of key radio algorithms, such as scheduling, beamforming and power optimisation – enhancing both spectral efficiency and energy performance, while also enabling advanced capabilities like predictive optimisation and radio sensing. This collaboration is an important step in our long‑term network strategy.”

Pallavi Mahajan, chief technology and AI officer at Nokia, added: “Through this collaboration with Orange, we are exploring how Nokia and Nvidia’s AI-RAN solution brings advanced AI and RAN functions together in a unified architecture. This will be instrumental in enabling the industry’s transition toward cognitive, AI‑native networks.” 

The agreement will be a big fillip for Nokia, which has found itself losing ground in the RAN market in recent years (especially in the US). The investment from, and partnership with, Nvidia appears to have given its RAN unit a new lease of life and the vendor will surely be using its expanded Orange relationship to get its AI-RAN foot in the door at other MNOs. 

This is also a big plus for Nvidia, which wants to get as many major MNOs on board with the idea of deploying GPU-based architecture in their radio access networks as quickly as possible, as other architectural options are emerging that can support RAN-based multifunctional processing. Among the alternatives are systems underpinned by Intel’s Xeon 6 system-on-chip (SoC), which is being put through its paces in RAN architectures by multiple operators – including Orange. Intel is also working closely with Ericsson on next-gen RAN system developments. 

- Ray Le Maistre, Editorial Director, TelecomTV

Email Newsletters

Sign up to receive TelecomTV's top news and videos, plus exclusive subscriber-only content direct to your inbox.

Subscribe

Cookies

TelecomTV uses cookies and third-party tools to provide functionality, personalise your visit, monitor and improve our content, and show relevant adverts.