6G news and analysis for the week ending Friday, 31st October, 2025.
The biggest story this week is unmistakable. NVIDIA announced a $1 billion strategic investment in Nokia alongside a partnership to build commercial-grade AI-RAN products for the 5G-to-6G transition. This is not a research collaboration or a memorandum of understanding — it is a capital commitment that fundamentally changes the dynamics of the RAN vendor landscape. With T-Mobile US signed up for field trials beginning in 2026 and Dell Technologies providing the server infrastructure, the partnership moves AI-RAN from conference-stage concept to commercial roadmap. NVIDIA’snew Aerial RAN Computer Pro platform is explicitly positioned as 6G-ready, designed for software-upgradeable evolution from 5G Advanced.
The investment thesis behind this move becomes clearer when viewed alongside Dell’Oro Group’s new 6G research report, which projects the RAN market will remain flat over the next decade until 6G capex begins ramping around 2030. With cumulative 6G RAN investments expected to account for 55 to 60 per cent of total RAN capex between 2029 and 2034, the prize is substantial for those positioned early.
Elsewhere, European collaboration produced a notable result: Ericsson, Nokia and Fraunhofer HHI jointly demonstrated a next-generation video codec with substantially higher compression efficiency than current standards, submitting their findings to the ITU-T and ISO/IEC standardisation bodies. With the next video codec standard expected to be operational from 2029-2030, this work is directly aligned with the 6G deployment timeline. And in Japan, SoftBank quietly advanced its Large Telecom Model into a domestically developed AI foundation, achieving over 90% accuracy in predicting communication quality during large-scale events — a practical demonstration of AI-driven network operations.
NVIDIA invests $1 billion in Nokia for AI-RAN
NVIDIA and Nokia announced a strategic partnership to develop commercial AI-RAN products, backed by a $1 billion NVIDIA investment in Nokia at a subscription price of $6.01 per share. The partnership will embed NVIDIA’s new Aerial RAN Computer Pro (ARC-Pro) platform at the heart of Nokia’s RAN portfolio. T-Mobile US will collaborate on field trials beginning in 2026, and Dell Technologies will provide PowerEdge servers to power the AI-RAN solution. Jensen Huang, founder and CEO of NVIDIA, said:
“Built on NVIDIA CUDA and AI, AI-RAN will revolutionize telecommunications — a generational platform shift that empowers the United States to regain global leadership in this vital infrastructure technology.”
Justin Hotard, President and CEO of Nokia, added:
“Our partnership with NVIDIA, and their investment in Nokia, will accelerate AI-RAN innovation to put an AI data center into everyone’s pocket.”
What’s new: NVIDIA introduced the Aerial RAN Computer Pro (ARC-Pro), described as a 6G-ready accelerated computing platform that combines connectivity, computing and sensing capabilities. The platform is designed to enable operators to move from 5G Advanced to 6G through software upgrades rather than hardware replacement. Nokia will accelerate the availability of its 5G and 6G RAN software on NVIDIA CUDA and expand its AirScale baseband architecture into the 6G era. The partnership also extends into AI networking, with Nokia’s SR Linux software for NVIDIA Spectrum-X Ethernet and exploration of Nokia’s optical technologies for future NVIDIA AI infrastructure.
Why this matters: This is the most significant financial commitment yet in the AI-RAN space. NVIDIA’s $1 billion investment in Nokia is not an R&D grant — it is a strategic equity stake that aligns incentives and signals long-term commitment. The announcement addresses the AI-RAN market, which Omdia estimates will represent a cumulative $200 billion opportunity by 2030. The involvement of T-Mobile US as first trial partner and Dell Technologies for server infrastructure creates an immediate commercial ecosystem. The framing is heavily US-centric, with repeated references to “America’s return to telecommunications leadership,” but the commercial implications are global. Nokia’s anyRAN approach, which allows new ARC-Pro cards to coexist with previously deployed AirScale baseband cards, offers operators an incremental migration path rather than a wholesale infrastructure replacement.
Implications for 6G standards: The ARC-Pro platform’s claim to be “6G-ready” rests on its software-defined architecture, which in principle allows new capabilities to be added through updates. This aligns with the broader industry consensus that 6G should be deployable on evolved infrastructure. However, the partnership raises questions about the relationship between vendor-specific AI-RAN implementations and the standardised interfaces that 3GPP will define in Release 20 and beyond. NVIDIA’s growing influence in the RAN ecosystem — now backed by a major equity position in one of the three largest RAN vendors — could shape how AI workload integration is specified in future standards.
View source material to read more
Samsung and SoftBank partner on AI-RAN research
Samsung Electronics and SoftBank Corp. signed a Memorandum of Understanding for joint research into next-generation communications technologies, including 6G and AI-RAN. The collaboration covers four candidate areas: 6G, AI for RAN, AI and RAN, and Large Telecom Models. The companies plan to identify new use cases, jointly develop core technologies and demonstrate their effectiveness. JinGuk Jeong, EVP and Head of Advanced Communications Research Center at Samsung Research, said:
“Through this collaboration with SoftBank, we aim to define meaningful use cases for both operators and end users, while securing key technologies for future commercialization.”
Hideyuki Tsukuda, EVP and CTO of SoftBank Corp., added:
“By combining our advanced expertise, we will accelerate the realization of next-generation networks that evolve to become more efficient and highly reliable through AI-RAN.”
What’s new: The MOU explicitly categorises next-generation communications into four distinct streams: pure 6G technology, AI applied to optimise RAN operations (AI for RAN), AI workloads integrated alongside RAN functions (AI and RAN), and Large Telecom Models. This taxonomy reflects the maturing industry understanding that AI-RAN is not a single concept but a spectrum of integration approaches. Samsung plans to host a Silicon Valley Future Wireless Summit in November to foster dialogue on AI-RAN research.
Why this matters: SoftBank is emerging as a key AI-RAN laboratory. This MOU with Samsung sits alongside SoftBank’s existing work on Large Telecom Models and its role in the AI-RAN Alliance. Samsung’s own AI-RAN Lab, established to accelerate prototyping and testing, demonstrated AI-RAN capabilities at MWC 2025. The four-pillar framework is analytically useful: it distinguishes between using AI to improve existing RAN functions, running AI services on RAN infrastructure, and applying generative AI foundation models to network operations. Each category carries different implications for network architecture and standardisation.
Implications for 6G standards: The distinction between “AI for RAN” and “AI and RAN” maps loosely onto the 3GPP Release 20 study items on AI/ML for RAN and the broader discussion about how 6G networks should host AI workloads. Samsung’s 3GPP 6G Workshop submission advocated for “sustainable user experience” and AI-native air interfaces. This collaboration provides a testbed for translating those principles into demonstrable technology.
View source material to read more
SoftBank evolves Large Telecom Model into domestic AI platform
SoftBank Corp. announced that it has evolved its Large Telecom Model (LTM) into a Japan-developed AI model by integrating “Sarashina,” a homegrown large language model developed by subsidiary SB Intuitions. The LTM enables end-to-end data processing entirely within Japan, using SoftBank’s proprietary network data and operational expertise. Internal deployment has begun, with the model achieving over 90% accuracy in predicting communication quality during large-scale events. Hideyuki Tsukuda, EVP and CTO of SoftBank Corp., said:
“The LTM enables AI-driven autonomous operations that reproduce the expertise and decision-making of experienced engineers, advancing base station optimization and improving communication quality.”
What’s new: SoftBank has moved the LTM from a general-purpose LLM foundation to a Japan-specific model capable of handling sensitive operational data within domestic data centres. The Communication Quality Prediction Model, fine-tuned from the LTM, was tested during the Kita City Fireworks Festival in September 2025, predicting connection volumes across more than 1,000 simulated base station conditions. SoftBank has also upgraded the LTM into a reasoning model incorporating Chain of Thought generation and reinforcement learning, enabling it to surpass traditional supervised learning approaches. A reinforcement learning environment using radio wave propagation simulators allows the model to continuously optimise radio conditions.
Why this matters: This is one of the most detailed public accounts of a telecom-specific AI foundation model in production use. While many operators talk about AI-driven network automation, SoftBank is demonstrating concrete capabilities: predicting quality degradation before events, optimising base station configurations and building autonomous learning loops. The data sovereignty dimension is significant — moving from English-centric general-purpose LLMs to a Japanese-language model trained and operated domestically addresses both compliance and contextual accuracy. The 90% prediction accuracy during a real-world stress event (a fireworks festival generating concentrated traffic) provides a credible benchmark.
Implications for 6G standards: Large Telecom Models represent a potential architectural building block for AI-native 6G networks. The 3GPP 6G Workshop heard multiple submissions advocating for standardised AI/ML frameworks, including the need for model lifecycle management and data collection standards. SoftBank’s work provides practical evidence of what these frameworks need to support. The question for standards bodies is whether LTM-style capabilities require new standardised interfaces or whether they can operate as vendor-specific implementations above the standards layer.
View source material to read more
Dell’Oro: 6G capex ramp to start around 2030
Dell’Oro Group published a new 6G Advanced Research Report projecting that the RAN market will remain flat over the next decade. The baseline scenario anticipates a 6G capex ramp dominated by Massive MIMO systems in the sub-7 GHz spectrum, utilising existing macro grids. Cumulative 6G RAN investments over the 2029-2034 period are projected to account for 55 to 60 per cent of total RAN capex. Stefan Pongratz, VP of RAN and Telecom Capex research at Dell’Oro Group, said:
“The RAN network plays a pivotal role in the broader telecom market. There are opportunities to expand the RAN beyond the traditional MBB use cases. At the same time, there are serious near-term risks tilted to the downside, particularly when considering the slowdown in data traffic.”
What’s new: The report confirms that worldwide RAN revenues have grown at just 1 per cent CAGR since 2000 in nominal USD terms, and the picture is worse when adjusted for inflation. The baseline 6G scenario assumes mobile networks will face utilisation challenges by end of decade, triggering the capex ramp. The report also outlines optimistic and pessimistic scenarios depending on mobile data traffic growth and the impact of use cases beyond traditional mobile broadband, including private wireless and fixed wireless access.
Why this matters: The flat market projection is a sobering backdrop for the 6G hype cycle. Operators are not in a growth phase — they are managing a mature RAN infrastructure with limited near-term investment incentives. The 55-60 per cent capex concentration in 6G between 2029 and 2034 suggests a sharp transition rather than a gradual ramp, which has implications for vendor revenue planning and operator budget cycles. The emphasis on sub-7 GHz Massive MIMO and existing macro grids points toward an evolutionary deployment model rather than the dense small-cell revolution that was promised for 5G millimetre-wave but largely did not materialise.
Implications for 6G standards: If the 6G capex ramp begins around 2030, this creates direct pressure on the 3GPP timeline. Release 20 study items need to produce actionable specifications by 2028 to give vendors sufficient lead time for product development and operator testing. The emphasis on sub-7 GHz spectrum reinforces the importance of the upper mid-band (7-8 GHz) frequency discussions that are central to current 3GPP radio interface studies and WRC-27 preparations.
View source material to read more
European trio drives 6G video coding standardisation
Ericsson, Nokia and Fraunhofer Heinrich Hertz Institute (HHI) have partnered to shape the next generation of video coding standardisation. The three organisations combined their video codec research expertise for the first time, demonstrating a new codec with considerably higher compression efficiency than current standards (H.264/AVC, H.265/HEVC and H.266/VVC) without significantly increasing complexity, while improving energy efficiency and scalability. Magnus Frodigh, Head of Research at Ericsson, said:
“Our partnership video codec achievement not only shows the ability of European technology leaders to come together and pioneer breakthroughs, but, by being prominent at the very start of the process, also flags our combined determination and commitment to shape the next generation of standards.”
What’s new: The partners submitted their joint work to the ITU-T Video Coding Experts Group and the ISO/IEC Moving Picture Experts Group (MPEG), where it was positively evaluated and is expected to help catalyse progress toward the next standardisation phase. The next video codec standard is expected to be operational from 2029-2030, aligned with forecast 6G deployment. Anticipated use cases include coding of professional, user and AI-generated content; gaming and 3D content; machine-consumption content; and automotive and industrial applications.
Why this matters: Video coding standards define the efficiency of content delivery across mobile networks for a decade or more. The current VVC standard already supports advanced streaming and immersive formats, but 6G’s anticipated use cases — holographic communication, real-time XR, AI-generated content — will demand substantially higher compression efficiency at lower latency and energy cost. The European collaboration is strategically significant: by aligning Ericsson, Nokia and Fraunhofer HHI early in the standardisation process, the partners are positioning Europe to influence a standard that will shape the digital media landscape through the 2030s.
Implications for 6G standards: While video codec standardisation runs through the ITU-T and ISO/IEC rather than 3GPP, the two processes are tightly coupled. The 6G radio interface must be designed to efficiently transport the content formats that the next video codec enables. The emphasis on energy efficiency and scalability in the new codec aligns with the sustainability goals that featured prominently across 3GPP 6G Workshop submissions.
View source material to read more
LG Electronics expands 6G and quantum research scope
LG Electronics hosted the “Future 2030 Summit” with KAIST and KRISS at its Seocho R&D Campus in Seoul. The event, which has been running as the “6G Grand Summit” since 2022, expanded its scope this year to include physical AI, quantum technology (Quantum-X) and space industry technology (Space Tech). LG Electronics demonstrated a new 6G full duplex radio (FDR) solution, a private 5G factory control system and a non-terrestrial network voice communication solution.
What’s new: The rebranding from “6G Grand Summit” to “Future 2030 Summit” reflects a broadening of LG Electronics’ research agenda beyond pure communications into adjacent emerging technologies. The FDR demonstration is noteworthy — full duplex radio, which allows simultaneous transmission and reception on the same frequency, could dramatically improve spectrum efficiency if the self-interference cancellation challenges can be solved at scale. The IoT-NTN voice communication solution enables calls in areas with poor terrestrial connectivity using satellite networks — a practical NTN demonstration aligned with the 3GPP NTN roadmap. The event also featured KAIST and KRISS demonstrations of AI semiconductor-based robots and quantum computing solutions.
Why this matters: LG Electronics is not a RAN equipment vendor, but it is a significant 3GPP contributor — its submission to the 6G Workshop advocated for AI-native air interfaces, ISAC, NTN integration and standalone architecture from day one. The company’s research partnerships with KAIST (through the jointly established 6G Research Center since 2019) and KRISS provide a pipeline from academic research to practical technology demonstrations. The expansion into physical AI and quantum research reflects the broader convergence of technologies that will characterise the 6G era. KRISS recently secured four Horizon Europe research projects, including 6G-relevant electromagnetic measurement work, strengthening the Korea-Europe research link.
Implications for 6G standards: Full duplex radio has been discussed in academic and industry research for years but faces significant implementation challenges. If LG Electronics can demonstrate viable FDR at commercially relevant power levels and bandwidths, it could influence the 3GPP Release 20 and 21 duplex technology discussions. LG Electronics’ 3GPP Workshop submission specifically called for “duplex technology evolution” as a RAN consideration for 6G.