Nokia’s AI-RAN Trials Expand to Eight Operators: The Long Road from Research to Commercial Reality

Nokia’s AI-RAN Trials Expand to Eight Operators: The Long Road from Research to Commercial Reality

On September 16, 2026, Nokia announced a significant expansion of its AI-RAN trials, naming eight operators—A1 Group, Chunghwa Telecom, du, e&, Mobily, stc, TPG Telecom, and Zain Saudi—that are advancing proofs of concept and live field trials across North America, Europe, Asia-Pacific, and the Middle East. The news, which sent Nokia’s stock up more than 3% in premarket trading, builds on earlier evaluations with T-Mobile, SoftBank, Indosat Ooredoo Hutchison, and NTT DOCOMO. It also follows Nokia’s July 2026 launch of what it calls the industry’s first commercial AI-native RAN platform, developed in collaboration with NVIDIA and its Aerial RAN Computer. Nokia reports that the platform has already delivered more than 20% improvements in spectral efficiency, with further gains expected through software updates in 2027 and 2028.

For an industry that has spent the better part of a decade debating the merits and feasibility of Open RAN, this announcement is more than a product update. It is a milestone in a longer historical arc—one that began with the promise of disaggregated, software-driven radio networks and has now arrived at the doorstep of AI-native commercial deployment. To understand why this moment matters, it helps to trace how we got here.

The Seeds of Disaggregation: From DAS to Open RAN

The story of AI-RAN begins with a much older idea: that radio access networks could be built from modular, interoperable components rather than monolithic, vendor-specific hardware. In the early 2010s, the telecom industry began experimenting with distributed antenna systems (DAS) and cloud RAN (C-RAN) architectures, which separated baseband processing from radio units. These early efforts were driven by a simple economic imperative: mobile operators wanted to break free from the proprietary interfaces that locked them into a single vendor’s ecosystem.

The formation of the O-RAN Alliance in 2018 gave this movement a formal structure. By defining open interfaces between radio units, distributed units, and centralized units, the alliance aimed to foster a multi-vendor market where operators could mix and match components from different suppliers. The vision was compelling: lower costs, faster innovation, and greater flexibility. But progress was slow. Interoperability testing proved complex, performance lagged behind integrated solutions, and the promised cost savings often failed to materialize. Skeptics pointed to the lack of mature software stacks and the difficulty of managing multi-vendor networks.

Despite these challenges, the Open RAN movement gained momentum. Governments, particularly in the United States and Europe, threw their weight behind it as a geopolitical tool to counter the dominance of Chinese vendors Huawei and ZTE. Operators like Rakuten Mobile in Japan and Dish Network in the United States built greenfield networks based on Open RAN principles, providing valuable real-world data. By the mid-2020s, Open RAN had moved from lab trials to early commercial deployments, though it remained a niche compared to traditional RAN.

The AI Inflection: When RAN Met Artificial Intelligence

As Open RAN matured, a new technological wave began to reshape the industry: artificial intelligence. Initially, AI was applied to network operations—predictive maintenance, anomaly detection, and energy optimization. But researchers soon realized that AI could also enhance the radio itself. By using machine learning algorithms to optimize beamforming, scheduling, and spectrum allocation, operators could squeeze more capacity out of existing spectrum. This was the birth of AI-RAN.

The concept gained traction in 2023 and 2024, with the formation of the AI-RAN Alliance—a group that now boasts 132 members and 33 demos. Early trials, such as those by NVIDIA and SoftBank, demonstrated impressive spectral efficiency gains. NVIDIA’s AI Aerial platform, for instance, achieved a 3x spectral efficiency improvement in SoftBank field trials. These results suggested that AI could be the killer app for Open RAN, providing the performance boost needed to justify the complexity of disaggregated networks.

But AI-RAN also raised new questions. Would it require specialized hardware, such as GPUs, that would undermine the goal of vendor neutrality? Could AI models be trained and deployed at scale without introducing unacceptable latency? And perhaps most importantly, would operators be willing to pay for yet another network upgrade cycle? These questions hung over the industry as the first AI-RAN platforms emerged.

Nokia’s Pivot: From anyRAN to AI-Native

Nokia, a stalwart of the traditional RAN market, had been somewhat late to the Open RAN party. But under the leadership of its Mobile Infrastructure division, the company pivoted aggressively. Its anyRAN software, launched in 2023, was designed to run on any cloud infrastructure, providing a bridge between traditional and open architectures. Then, in October 2025, Nokia announced a strategic partnership with NVIDIA, which included a $1 billion investment giving NVIDIA a 2.9% stake in the company. The partnership aimed to combine Nokia’s RAN software with NVIDIA’s Aerial RAN Computer, integrating AI processing directly into the radio access network.

In July 2026, Nokia launched its commercial AI-native RAN platform, which it described as the industry’s first. The platform promised not only spectral efficiency gains but also a path from 5G to 6G, turning radio networks into programmable platforms that could support both connectivity and AI workloads. At the launch, NVIDIA CEO Jensen Huang called RAN “the next AI infrastructure” and envisioned a future where telecom networks become distributed computers that run AI applications at the edge. Nokia’s platform, he said, would turn RAN into a “planet-scale AI computer.”

The September 2026 announcement is the first major update since that launch. It expands the roster of operators engaged in AI-RAN trials from a handful to eight named players, plus previously announced partners. While the 20% spectral efficiency improvement was already disclosed in July, the broader operator engagement signals that AI-RAN is moving from early evaluation to lab and live field trials—a critical step on the path to commercial deployment.

The Eight Operators: A Geographically Diverse Funnel

The eight operators named in the announcement span four regions, reflecting a broad-based interest in AI-RAN. A1 Group operates in Austria and Central and Eastern Europe; Chunghwa Telecom is Taiwan’s largest integrated telecom operator; du and e& are major players in the United Arab Emirates; Mobily, stc, and Zain Saudi are leading Saudi Arabian operators; and TPG Telecom is a major Australian provider. This geographic diversity is notable because it suggests that AI-RAN is not a niche interest limited to early adopters in a single market. Instead, it is attracting operators from both developed and emerging markets, each with different spectrum portfolios, network densities, and business models.

For these operators, the appeal of AI-RAN is straightforward: the ability to extract more capacity from existing spectrum without deploying new hardware. Nokia’s platform, which combines its anyRAN software with NVIDIA’s Aerial RAN Computer, is designed to be deployed as a software upgrade on existing 5G networks, with a path to 6G. This software-first approach is a departure from the traditional hardware refresh cycle that has long characterized the RAN market. It promises to reduce capital expenditure and accelerate time-to-market for new services.

However, the announcement also leaves many questions unanswered. Nokia did not provide detailed results for each participating operator, nor did it specify when the trials might lead to commercial deployments. The financial contribution of AI-RAN to Nokia’s bottom line remains undisclosed. These omissions are not unusual for early-stage trials, but they highlight that the commercial case for AI-RAN is still unproven.

The 20% Gain: A Milestone, Not a Finish Line

The more than 20% improvement in spectral efficiency is a significant achievement. Spectral efficiency is a measure of how much data can be transmitted over a given amount of spectrum, and it is a key metric for operators looking to maximize the value of their spectrum assets. A 20% gain could translate into substantial capacity improvements, allowing operators to support more users and higher data rates without acquiring additional spectrum.

But it is important to put this figure in context. Nokia had already disclosed a greater-than-20% efficiency result when it launched the platform in July. The September announcement reaffirms that result but does not provide new performance data. Moreover, Nokia’s roadmap targets a greater-than-100% spectral efficiency improvement by 2028, which means the current gain is only the first step. The company expects further improvements through software updates in 2027 and 2028, but these are projections, not guarantees.

Industry analysts have been cautious. Remy Pascal, Practice Leader for Mobile Infrastructure at Omdia, noted that the breadth of operator engagement shows AI-RAN is evolving from a research topic into a strategic priority. But he also implied that commercial deployment schedules and financial contributions remain unclear. The lack of disclosed revenue figures is a reminder that trials and proofs of concept do not necessarily translate into near-term sales.

The Competitive Landscape: Ericsson, Huawei, and the Rest

Nokia’s announcement also invites comparison with its competitors. Ericsson, Nokia’s primary rival in the RAN market, has been developing its own AI-RAN capabilities. In 2025, Ericsson launched an AI-in-RAN subscription service and partnered with SK Telecom on an AI-RAN pilot. However, Ericsson’s stock barely moved on the day of Nokia’s announcement, suggesting that the market does not yet view AI-RAN as a category-wide rerating event. As one analysis noted, “a genuine industry rerating around AI-RAN would lift both names together rather than Nokia alone.”

Huawei, meanwhile, has been pushing its own AI-RAN vision with SingleRAN 22.1, which it describes as a “mobile AI foundation.” But Huawei’s ability to compete in Western markets is constrained by geopolitical factors, leaving Nokia and Ericsson as the primary beneficiaries of operator interest in AI-RAN outside China.

The competitive dynamics extend beyond traditional RAN vendors. NVIDIA’s entry into the RAN market through its partnership with Nokia represents a new kind of competition—one that pits GPU-accelerated computing against specialized baseband silicon. NVIDIA’s Aerial RAN Computer is designed to run AI workloads alongside RAN processing, potentially turning every cell site into a mini data center. This vision aligns with NVIDIA’s broader strategy of extending its CUDA platform from data centers into networking, vehicles, robotics, and edge computing. If successful, it could fundamentally reshape the RAN value chain.

The Road Ahead: Pilots, Commercialization, and 6G

Nokia has said it expects pilot deployments to begin late in 2026, with commercial availability following in 2027 through a software-subscription model. This timeline is ambitious but not unrealistic, given the progress demonstrated so far. However, several challenges remain.

First, the business model is unproven. Operators are accustomed to paying for hardware, not software subscriptions. Convincing them to shift to a software-centric model will require clear evidence of return on investment. The 20% spectral efficiency gain is a start, but operators will want to see how it translates into revenue or cost savings.

Second, the technical integration of AI into RAN is complex. Running AI workloads alongside real-time RAN processing requires careful orchestration to avoid latency and reliability issues. Nokia’s partnership with NVIDIA is designed to address this, but real-world performance across diverse network environments remains to be proven.

Third, the path to 6G is still being defined. AI-native 6G networks are expected to be highly programmable and intelligent, but standards are not yet finalized. Nokia’s claim that its platform provides a software upgrade path to 6G is forward-looking, but it depends on the industry coalescing around common architectures.

Finally, the competitive landscape is evolving rapidly. Ericsson, Huawei, Samsung, and a host of startups are all investing in AI-RAN. Nokia’s early lead in commercializing an AI-native platform may be challenged as rivals bring their own solutions to market.

Conclusion: A Milestone in a Longer Journey

Nokia’s expanded AI-RAN trials are a significant milestone in the long journey from Open RAN research to commercial reality. They demonstrate that the industry’s largest vendors and operators are taking AI-RAN seriously, and they provide a foundation for the next phase of development. But they also highlight how much work remains. The 20% spectral efficiency gain is promising, but it is not yet revenue. The eight operators are engaged, but they have not committed to commercial deployments. The platform is available, but the business model is still being tested.

For an industry that has often been characterized by hype cycles and unmet expectations, this cautious progress is perhaps a healthy sign. AI-RAN is not a silver bullet, but it may be the most promising path to making Open RAN economically viable at scale. As Nokia’s Emma Falck put it, the growing ecosystem of partners across regions demonstrates that AI-RAN “is becoming a shared industry priority, enabling more intelligent, efficient and software-driven networks.” If that priority translates into commercial deployments in 2027 and beyond, the decade-long journey toward disaggregated, intelligent radio networks will have reached its most important destination yet.

Sources

Related Posts

Nokia's AI-RAN Field Trials Go Global: 5 Things Operators Need to Know

Nokia expands AI-RAN trials to eight operators across four regions, reports 20% spectral efficiency gains and plans software roadmap through 2028.

NTT Docomo's Multi-Vendor Open RAN Goes Commercial: A Skeptical Executive's Q&A with Nokia

NTT Docomo has commercially deployed Nokia's multi-vendor Open RAN base station. We answer the tough questions skeptical operators are asking.