Nokia’s AI-RAN Platform Puts a Number on the Software-Defined RAN Bet

Nokia's AI-RAN Platform Puts a Number on the Software-Defined RAN Bet

Analyst(s): Nick Patience
Publication Date: July 15, 2026

Nokia has launched what it calls the industry’s first commercial AI-RAN platform, built on its anyRAN software and NVIDIA’s Aerial AI-RAN platform. The company is targeting more than 100% spectral efficiency gains by 2028, delivered through three hardware deployment paths and a new software subscription model.

What Is Covered in This Article:

  • Nokia announced the industry’s first commercial AI-RAN platform on July 15, 2026, built on its anyRAN software and NVIDIA’s Aerial AI-RAN platform, targeting more than 100% spectral efficiency gains by 2028.
  • The platform is delivered across three hardware paths: a GPU-powered AirScale capacity plug-in for existing deployments, a standalone AI-RAN node, and GPU-powered COTS server options for cloud-native operators.
  • Nokia is moving from custom silicon to NVIDIA’s merchant silicon platform and will monetize new AI-RAN capabilities through a value-based software subscription model, with pricing details still to come.
  • T-Mobile US is the only operator with confirmed dates on Nokia’s public deployment timeline (field trial in 2026, commercial start in 2027), though Nokia counts Vodafone, BT, Orange, Deutsche Telekom, Elisa, SoftBank, NTT DOCOMO, and Indosat Ooredoo Hutchison among its broader AI-RAN customer engagements.
  • Nokia’s claimed spectral efficiency gains are roughly five to ten times the scale of what rival Ericsson has already put into commercial deployment, a gap Nokia executives attribute to more compute-intensive algorithms rather than measurement differences.

The News: Nokia has announced what it calls the industry’s first commercial AI-RAN platform, built on the company’s AI-native anyRAN software and NVIDIA’s AI Aerial AI-RAN platform. According to Nokia, the platform will deliver more than 100% spectral efficiency gains by 2028, effectively doubling the capacity operators can extract from the spectrum they already own, with an interim target of 50% gains by 2027, building on more than 20% gains Nokia says it has already demonstrated through AI-driven radio innovations.

The platform is delivered across three hardware options: a GPU-powered AirScale capacity plug-in module for operators with an existing Nokia installed base, a standalone GPU-powered AI-RAN node for greenfield or high-capacity deployments and GPU-powered COTS server options for operators pursuing cloud-native architectures, with additional merchant silicon support from Marvell. All three run on a single anyRAN software stack and are fully Open RAN compliant. New AI-RAN capabilities will reach customers through a subscription-based commercial model, which Nokia says delivers improved total cost of ownership and performance at no hardware premium relative to existing baseband systems.

Nokia’s AI-RAN solutions are scheduled to enter pilot deployments by the end of 2026 and become commercially available in 2027. T-Mobile US is named as the lead customer on Nokia’s public timeline, while Vodafone, BT, Orange, Deutsche Telekom, Elisa, SoftBank, NTT DOCOMO, and Indosat Ooredoo Hutchison are named as part of Nokia’s broader AI-RAN customer engagement.

Nokia’s AI-RAN Platform Puts a Number on the Software-Defined RAN Bet

Analyst Take: Nokia’s announcement comes with the kind of top-line number that invites scrutiny: more than 100% spectral efficiency gains by 2028, delivered without a hardware premium, on a platform Nokia is positioning as the industry’s first commercial AI-RAN offering.

The 2x claim is a target for 2028, built from staged software releases: roughly 20% already demonstrated, 50% targeted for 2027, and more than 100% by 2028. Nokia executives were candid on the pre-briefing call that the eventual gain will vary by fronthaul split, radio channel conditions, mobility and cell-edge behavior, and that the biggest benefit accrues to dense, highly loaded cells rather than the network as a whole. None of that invalidates the target, but it does mean the number is currently a roadmap target.

The gap becomes more interesting when held up against what Ericsson has already put into commercial service. Ericsson’s AI in RAN software, live since June 2026 across more than 15 deployments, including SoftBank, Bell, SK Telecom, and Rogers, claims roughly 10% spectral efficiency and 20% downlink throughput gains today, without GPUs. Nokia’s answer, when pressed directly on the disparity during the briefing, was that its platform can run substantially more compute-intensive algorithms, including deep receivers and non-linear channel estimation techniques, than a fixed-function ASIC can support. That is a coherent technical argument. It also underscores the difference in kind: Ericsson’s gains are already shipping, while Nokia’s is a multi-year target contingent on algorithms still being developed with NVIDIA.

One Lead Customer and a Business Model Taking Shape

Nokia lists nine operators as engaged on AI-RAN: T-Mobile US, Vodafone, BT, Orange, Deutsche Telekom, Elisa, SoftBank, NTT DOCOMO and Indosat Ooredoo Hutchison, with T-Mobile US named at every milestone: 2026 field trial, 2027 commercial start, 2028 mass deployment. That’s a reasonable go-to-market sequence for a first commercial platform, but it means the near-term proof points analysts should watch for will come from a single operator relationship, not the broader roster of operators Nokia has named as engaged.

The commercial model is similarly under-specified. Nokia confirmed on the briefing call that new AI-RAN capabilities will be sold through a subscription, described repeatedly as value-based rather than tied to a fixed per-unit price, with further detail promised in the coming months. That is a meaningful shift for an industry accustomed to buying baseband hardware outright, and it puts Nokia’s AI-RAN economics closer to a consumption or outcome-based software model than a traditional capex refresh. Operators will reasonably want to know whether value-based means priced to the spectral efficiency delivered, the capacity unlocked, or something else entirely, before committing.

GPUs part of, but not the story

Executives on Nokia’s briefing on the announcement said that GPU-based compute is technically necessary for AI-native RAN (parallelism across users, layers, antennas, and beams; on-site fine-tuning; tensor-heavy sensing workloads) but that this is not fundamentally a GPU story. Both are true in a narrow sense. The near-term capacity roadmap running through 2027-2028 is substantially a software release cadence; the deeper GPU dependency shows up further out, in the 6G, integrated sensing, and physical AI vision Nokia is building toward with NVIDIA’s AI Aerial platform.

That distinction matters commercially. Nokia’s own messaging makes it clear that no hardware premium doesn’t mean no hardware; it just means the GPU cost is not broken out as a separate line item, likely absorbed into the subscription price rather than eliminated. Operators evaluating the total cost of ownership will want that unpacked before taking the no-premium claim at face value.

The Real Differentiator May Be the Optionality, Not the Number

Where Nokia’s announcement is on firmer ground is the three-path deployment structure: a plug-in module for the existing AirScale install base, a standalone AI-RAN node for greenfield or high-capacity sites and cloud-native COTS options for operators building centralized or distributed architectures, all on one software stack, with no forklift replacement required. That flexibility is an answer to operators’ understandable reluctance to commit to a single hardware path this early in the AI-RAN cycle, and it’s arguably a stronger differentiator than the efficiency number, precisely because it’s already defined rather than promised.

What This Means for the Market

Nokia’s move sharpens what was already the industry’s most consequential AI-RAN debate: whether custom silicon or merchant silicon wins the next RAN architecture cycle. Nokia has picked its side decisively, backed by a $1 billion NVIDIA equity stake last October and NVIDIA’s own pitch of RAN as a planet-scale AI computer. Ericsson, betting on purpose-built silicon and a faster time-to-shipping software story, is making the opposite bet. Both companies are chasing the same operator capex, and both are relying on claims that won’t be fully testable in the field for another one to two years. For operators, the practical calculus is less about which vendor’s spectral efficiency number is bigger and more about which platform’s economics, openness, and 6G upgrade path they trust furthest out. For the rest of the industry, Nokia’s launch is the clearest signal yet that AI-RAN has moved from an industry talking point to a genuine architectural fork, one that will likely take until 2028 to resolve on the evidence rather than the roadmap.

What to Watch:

  • Whether Nokia’s 50% spectral efficiency target for 2027 lands on schedule, and on which operator networks it is first measured and disclosed.
  • Pricing detail on Nokia’s AI-RAN software subscription model and how operators react to a consumption or value-based structure versus traditional hardware procurement.
  • Product specifics and capacity figures for Nokia’s standalone AI-RAN node, one of three deployment paths still without published performance numbers.
  • Whether any of Nokia’s other named operators (Vodafone, BT, Orange, Deutsche Telekom, Elisa, SoftBank, NTT DOCOMO, Indosat Ooredoo Hutchison) move onto a public deployment timeline alongside T-Mobile US.
  • How Ericsson, Samsung, and Huawei respond competitively and whether the AI-RAN Alliance and O-RAN Alliance formalize interfaces like Nokia’s proposed E3 specification industry-wide, rather than vendor by vendor.

See the press release on Nokia’s AI-RAN platform launch on the Nokia website.

Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.

Other Insights From Futurum:

Can Nokia’s Gemini-Based Network Agents Make Autonomous Networks Practical?

Is Telefónica and Nokia’s Agentic AI Push a Catalyst for Network API Adoption?

Author Information

Nick Patience is VP and Practice Lead for AI Platforms at The Futurum Group. Nick is a thought leader on AI development, deployment, and adoption - an area he has researched for 25 years. Before Futurum, Nick was a Managing Analyst with S&P Global Market Intelligence, responsible for 451 Research’s coverage of Data, AI, Analytics, Information Security, and Risk. Nick became part of S&P Global through its 2019 acquisition of 451 Research, a pioneering analyst firm that Nick co-founded in 1999. He is a sought-after speaker and advisor, known for his expertise in the drivers of AI adoption, industry use cases, and the infrastructure behind its development and deployment. Nick also spent three years as a product marketing lead at Recommind (now part of OpenText), a machine learning-driven eDiscovery software company. Nick is based in London.

Related Insights
The Active Storage Revolution: VAST and Cloudera Team Up to Cure Enterprise GPU Starvation
July 15, 2026

The Active Storage Revolution: VAST and Cloudera Team Up to Cure Enterprise GPU Starvation

Brad Shimmin, VP and Practice Lead at Futurum, explores the new strategic partnership between VAST Data and Cloudera. By integrating the VAST AI OS with Cloudera data services, the vendors...
Intel Pours €5 Billion into Ireland to Feed the Xeon Surge
July 14, 2026

Intel Pours €5 Billion into Ireland to Feed the Xeon Surge

Brendan Burke, Research Director at Futurum, examines how Intel's €5 billion investment in Ireland to expand Xeon 6 manufacturing signals confidence in its foundry business and commitment to meeting surging...
FOXTRON's Adoption of Dimensity AX C-X1 Validates MediaTek's Automotive Ambitions
July 13, 2026

FOXTRON’s Adoption of Dimensity AX C-X1 Validates MediaTek’s Automotive Ambitions

Olivier Blanchard, Research Director at The Futurum Group, examines how FOXTRON's adoption of MediaTek's Dimensity AX C-X1 platform moves AI-defined vehicle ambitions from platform development into commercial automotive deployment....
Adobe CX Enterprise CoWorker is Ready. Is the Enterprise?
July 13, 2026

Adobe CX Enterprise CoWorker is Ready. Is the Enterprise?

Nick Patience, VP & Practice Lead for AI Platforms at The Futurum Group, examines Adobe's CX Enterprise Coworker bet at Summit London, and the gap between the vision and where...
Can UST and Claude Make Physical AI the Next Enterprise Standard?
July 11, 2026

Can UST and Claude Make Physical AI the Next Enterprise Standard?

UST integrated Anthropic's Claude into core engineering platforms, reducing chip validation cycles by 50-70% and training 20,000 engineers on AI-native operations across multiple sectors....
Are Podcasts the New Playbook for Engineering Leaders Work through AI in the SDLC?
July 11, 2026

Are Podcasts the New Playbook for Engineering Leaders Work through AI in the SDLC?

Qodo releases a curated podcast guide for engineering leaders navigating the rapidly evolving AI platforms market. With agentic AI adoption accelerating, leaders need practical resources to build AI literacy and...

Book a Demo

Welcome

The vision behind everything in Futurum’s Custom Research practice is this: research should show you what is happening, what comes next, and what to do about it. It should be personal to each audience, easy for people to grasp, and structured so LLMs can reason over it accurately. And it should be fast and turnkey; you want answers now, not another project to carry for quarters.

Whether you are defining business, channel, or go-to-market strategy; evaluating vendors or justifying ROI; or commissioning research to fill an emerging market need, we have your back, with a program that answers your questions with the objectivity and credibility to drive real decisions.

To do it, we bring unmatched data to bear: Futurum research, surveys, and market projections; validated market feeds; ETR’s 15 years of insight from 10,000 technology decision-makers; G2’s buyer and user data; and what our analysts hear every day. Add leading primary collection, from AI-moderated voice interviews to surveys and analyst-led interviews, all turnkey, and every project comes out credible, nuanced, and actionable.

And we don’t just drop the results in your lap. For internal work, we provide analyst-led sessions, interactive dashboards, and a range of formats. For market-facing work, Futurum delivers turnkey activation and amplification that actually gets seen, by people and by LLMs, through our media and share of voice. This is research that moves decisions and markets.

We will meet you wherever you are, from a fast-turn brief to a multi-year program, and shape the work to your goals, timeline, and budget. The right program for your moment.

If any of this is useful, I would love to talk.

Benjamin Brown, VP Custom Research, Futurum Research

Benjamin Brown

VP, Custom Research · The Futurum Group

Newsletter Sign-up Form

Get important insights straight to your inbox, receive first looks at eBooks, exclusive event invitations, custom content, and more. We promise not to spam you or sell your name to anyone. You can always unsubscribe at any time.

All fields are required






Thank you, we received your request, a member of our team will be in contact with you.