6G news and analysis for the week ending Friday, 20th March, 2026.
This week’s theme is the accelerating convergence of AI and telecoms infrastructure. NVIDIA used its GTC 2026 conference to unveil “AI grids”, a concept that transforms telecom operators’ distributed network sites into geographically-dispersed AI inference platforms. Six major operators, including AT&T, Comcast, T-Mobile and Indonesia’s Indosat, announced concrete deployments, moving beyond the MWC 2026 commitments covered two weeks ago into operational infrastructure. Ericsson and SK Telecom reinforced the AI-RAN trajectory by signing a six-year MoU spanning joint R&D through to potential commercialisation by 2031.
On the geopolitical front, the NTIA made two significant moves. It pivoted its Innovation Fund away from Open RAN toward AI-native network architecture, aligning federal funding with the Trump Administration’s AI Action Plan. Separately, NTIA administrator Arielle Roth announced plans for a “6G Call for Action” designed to unite US allies around a software-defined, open-interface model for 6G development. Both moves signal that Washington views 6G as a strategic technology competition.
Elsewhere, ETSI released the first version of its open-source Operator Platform for telco cloud federation and 6G experimentation. The Netherlands secured €142 million from its National Growth Fund for Phase 2 of the Future Network Services programme. And e& UAE partnered with Khalifa University to publish an 80-page AI-native 6G architecture whitepaper introducing a dedicated “AI plane” alongside conventional network planes.
NVIDIA and operators build distributed AI grids
NVIDIA announced at GTC 2026 that leading US and Asian operators are building AI grids to power and monetise AI inference services at the edge. It describes these AI grids as geographically distributed AI infrastructure embedded across network footprints. The initiative extends NVIDIA’s telecoms strategy from the AI-native 6G platform commitment announced two weeks ago at MWC 2026 into actual distributed computing deployments. AT&T is partnering with Cisco and NVIDIA to build an AI grid for IoT, running inference closer to data sources for mission-critical applications. Shawn Hakl, SVP of Product at AT&T Business, said:
“Scaling AI services that are both highly secure and accessible for enterprises and developers is a core pillar of our IoT connectivity strategy.”
In addition, Comcast is developing its broadband footprint into an AI grid for real-time, hyper-personalised experiences, working with NVIDIA, Decart, Personal AI and HPE. Spectrum (Charter) is deploying AI infrastructure across more than 1,000 edge data centres. T-Mobile is exploring edge AI applications using NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, with developers piloting smart-city, industrial and retail applications. Indosat Ooredoo Hutchison is connecting its sovereign AI factory with distributed edge and AI-RAN sites across Indonesia.
What’s new: AI grids convert existing telecom real estate, power and connectivity into a distributed computing platform for AI inference. NVIDIA estimates the global telecom sector has approximately 100,000 distributed network data centres with enough spare power to offer more than 100 gigawatts of new AI capacity. The AI Grid Reference Design defines building blocks for deploying and orchestrating AI across distributed sites. Partners including Cisco, HPE, Armada, Rafay and Spectro Cloud are building the orchestration and control plane layers. Akamai is expanding its Inference Cloud across more than 4,400 edge locations with NVIDIA GPUs.
Why this matters: This represents a structural shift in how AI is delivered. Rather than centralised hyperscale data centres handling all inference, AI grids position telecom networks at the centre of distributed AI delivery. The economics look interesting, as operators have the real estate, power and low-latency connectivity that AI inference demands. For the telecoms industry, this offers a monetisation route beyond connectivity. Masum Mir, SVP at Cisco, framed it as enabling operators to “participate in the AI value chain.” The AI-RAN dimension is also significant: T-Mobile and Indosat are both exploring how cell sites can support AI workloads alongside advanced 5G connectivity, validating the AI-RAN architecture that NVIDIA has been championing.
Implications for 6G standards: AI grids reinforce the industry trajectory toward 6G as a compute-and-connectivity platform rather than a pure communications upgrade. NVIDIA’s 3GPP 6G Workshop submission in 2025 argued that operators should move beyond “bit transport” to become compute and data platforms. These deployments create the operational foundation for that vision. As 3GPP Release 21 takes shape, the question of how AI workload orchestration interfaces with RAN and core network functions will become increasingly important.
View source material to read more
Ericsson and SK Telecom deepen AI-RAN collaboration to 2031
Ericsson and SK Telecom signed a Memorandum of Understanding to strengthen collaboration on mobile network technologies spanning 5G to 6G. The agreement runs until March 2031 and covers joint R&D, trials and potential commercialisation across AI-powered RAN, 5G monetisation, open and autonomous networks, zero trust security and 6G standardisation. Mårten Lerner, Head of Networks Strategy and Product Management at Ericsson, said:
“We’re pleased to deepen our partnership with SK Telecom to accelerate AI-powered RAN, 5G monetisation, open autonomous networks and zero-trust security, while laying the groundwork for AI-native 6G.”
Yu Takki, Head of Network Technology Office at SK Telecom, added:
“Our collaboration with Ericsson will be a core engine propelling the evolution of AI-powered networks, paving the way toward 6G.”
What’s new: The collaboration targets five technology domains: AI-powered RAN to enhance performance and energy efficiency; 5G advancements for new business opportunities; open and autonomous networks for multi-vendor environments; end-to-end zero trust security across 5G Advanced and 6G; and 6G standardisation including spectrum strategy, extreme MIMO evolution, energy efficiency and Integrated Sensing and Communication (ISAC). The 2031 timeline explicitly encompasses potential 6G commercialisation.
Why this matters: SK Telecom published its comprehensive ATHENA 6G architecture white paper just three weeks ago, detailing a domain-by-domain roadmap built on AI-native RAN, evolved core and data monetisation. This MoU with Ericsson provides the vendor partnership to execute that vision. SKT was recently elected to the board of the AI-RAN Alliance and has demonstrated GPU-based L1 processing. Ericsson completed the world’s first pre-standard 6G over-the-air trial in Texas earlier this month. The combination creates a potent research-to-deployment pipeline. The zero trust emphasis is also notable, following the GCOT 6G Security and Resilience Principles launched at MWC with endorsement from both companies.
Implications for 6G standards: Both companies are active 3GPP contributors with aligned positions on AI-native architecture, standalone deployment from day one, and realistic performance targets. Their joint focus on ISAC and extreme MIMO evolution maps directly onto active Release 20 study items. The explicit standardisation dimension of this MoU suggests coordinated contributions to 3GPP in these areas.
View source material to read more
NTIA pivots Innovation Fund to AI-RAN and plans 6G ally coalition
The US National Telecommunications and Information Administration (NTIA) made two significant policy announcements this week. First, it redirected the Public Wireless Supply Chain Innovation Fund away from Open RAN and toward AI-native network architecture, cancelling pending awards under its third funding round. Second, NTIA administrator Arielle Roth announced plans for a “6G Call for Action” to unite US allies around a shared approach to next-generation wireless.
On the Innovation Fund, NTIA said the new direction will focus on integrating AI into Radio Access Networks, supporting the Trump Administration’s AI Action Plan and the Executive Order on promoting exports of American AI technology. A public listening session is scheduled for 23rd March. Roth said:
“The telecom market is global in nature, and in order for our domestic innovators to achieve scale, it’s important to rally our allies around where America shines.”
On the 6G initiative, speaking at a sponsored Politico event on 18th March, Roth described the effort as a pledge to align allies on standards, R&D, market access, security, resilience and spectrum policy. She emphasised a software-defined model for 6G as opposed to what she described as a hardware-driven legacy approach. “The way that America and many of our western allies are trending, and where we shine, is in more of a software-based model of continuous innovation,” Roth said.
What’s new: The Innovation Fund pivot is a tangible policy shift. Launched in 2023 to advance US leadership in wireless technology and supply chain resilience, the fund was originally focused on Open RAN. Its reorientation toward AI-RAN reflects how rapidly the industry’s centre of gravity has moved. The “6G Call for Action” is a new geopolitical initiative that goes beyond previous US 6G positioning, explicitly seeking allied commitments around software-defined networks, open interfaces, edge computing and AI. No launch date has been specified.
Why this matters: These announcements represent the most concrete US federal policy moves on 6G this year. The Innovation Fund redirect from Open RAN to AI-RAN mirrors the broader industry shift visible in NVIDIA’s AI grid deployments and the AI-RAN Alliance’s growth to over 130 members. The “Call for Action” framing positions 6G as a strategic technology competition, with Roth drawing an explicit contrast between software-based and hardware-driven development models. The emphasis on rallying allies early in the standards process reflects Washington’s determination to shape 6G’s direction before specifications are locked.
Implications for 6G standards: US policy alignment around AI-native, software-defined architectures directly reinforces the 3GPP consensus on AI-native design principles for 6G. However, the geopolitical framing introduces risk: if the “Call for Action” is perceived as an attempt to predetermine standards outcomes in favour of specific national industry players, it could create friction in the multilateral 3GPP process. The cancellation of Open RAN funding in favour of AI-RAN also raises questions about the future of the original Open RAN supply chain diversification objectives.
View source material to read more — Innovation Fund
View source material to read more — 6G Call for Action
ETSI launches open-source operator platform for 6G experimentation
ETSI released the first version of OpenOP (Open Operator Platform), an open-source platform designed to enable telco cloud network federation and 6G experimentation. The platform provides a framework for telecom network capabilities to be consumed by vertical application providers and developers through CAMARA APIs.
OpenOP Release 1 includes an Open Exposure Gateway for CAMARA API exposure, a Service Resource Manager for application lifecycle management, a Federation Manager for cross-domain resource discovery, and an AI Integration module that uses large language models to translate natural language prompts into CAMARA API operations. The platform has been validated through integration with an open-source 3GPP Network Exposure Function for Location and Quality on Demand APIs.
What’s new: This is the first fully integrated open-source Operator Platform release from ETSI. The SDG OpenOP group was established only months ago and has already grown to 24 members including BT, Orange Romania, OTE, Motorola, Fraunhofer and the University of Bristol. Eight EU-funded 6G projects are contributing to the initiative. The AI Integration module is particularly notable, exposing a Model Context Protocol (MCP) server that enables intent-based interaction with telecom network capabilities through LLMs.
Why this matters: API exposure and developer-friendly access to network capabilities are widely viewed as essential for 6G monetisation. The CAMARA APIs, backed by the GSMA and the Linux Foundation, represent the leading industry effort to standardise these interfaces. OpenOP provides an open-source implementation that researchers and developers can use to experiment with 5G-Advanced and 6G network capabilities without proprietary infrastructure. The federation capability is important for multi-operator environments and cross-border experimentation, which are critical for validating 6G concepts at scale.
Implications for 6G standards: OpenOP follows GSMA Operator Platform Group specifications and integrates with 3GPP network exposure mechanisms. As 6G standards take shape, open-source reference implementations like this can accelerate the feedback loop between specification development and practical validation.
View source material to read more
Netherlands secures €142 million for 6G
The Future Network Services (FNS) consortium in the Netherlands received €142 million from the Dutch National Growth Fund for Phase 2 of its 6G development programme. The funding enables the consortium of more than 60 companies and knowledge institutions to move from research into concrete 6G technology, products and services. Tjark Tjin-A-Tsoi, CEO of TNO (Netherlands Organisation for Applied Scientific Research), said:
“With this investment in 6G, TNO is working with more than 60 companies and knowledge institutions to build the backbone of our future digital economy.”
What’s new: Phase 1 produced several notable results, including a 4.6-kilometre wireless optical link at a world record speed of 5.7 terabits per second and NXP’s first Wi-Int chips based on a new international standard. The consortium built Oakestra, an AI-driven tool for autonomous mobile network management, and established the National 6G Testbed across five regional locations. Phase 2 shifts focus to commercial value creation, with more than 100 SMEs signed up for pilots. A 6G Business Board will manage innovation priorities. International partnerships with Finland, Sweden, Germany, Japan and Taiwan will be intensified.
Why this matters: The Netherlands has a strong position in the 6G supply chain through companies like NXP and research organisations including TNO and TU Eindhoven. The Phase 2 emphasis on SME participation and commercial piloting addresses a frequent criticism of national 6G programmes: that they produce research outputs but struggle to translate them into viable products. The stated ambition to develop the first fully wireless factory in the Netherlands provides a concrete vertical use case target.
Implications for 6G standards: FNS partners are active in 3GPP through Dutch operator KPN and research organisations. The programme’s focus areas, including AI-driven network management, sensing and cloud-native infrastructure, align with the Release 20 and 21 study items. The five-site National 6G Testbed provides infrastructure for validating technologies before they enter the standards process.
View source material to read more
UAE partners propose AI-native 6G architecture
e& UAE and Khalifa University have published a whitepaper titled “6G AI-Native Networks: Architecture, Intelligence, and the Path to Autonomous Connectivity.” The 80-page document proposes a dedicated “AI plane” integrated alongside user, control and management planes as the foundation for what the partners describe as “true AI-native” 6G. Marwan Bin Shakar, CTO at e& UAE, stated:
“6G is not about incremental performance; it is about redefining the network as an intelligent system.”
What’s new: The whitepaper introduces five enabling pillars for AI-native 6G: pervasive AI/ML frameworks, distributed cloud-edge computing, advanced technologies including integrated sensing, open programmable architectures and sustainability-driven design. It defines measurable AI-native KPIs including decision latency for closed-loop autonomy, learning accuracy and energy efficiency per AI inference. The paper proposes standardised cross-domain interfaces for multi-vendor intelligence at scale and outlines a framework for distributed AI agents, AI lifecycle orchestration and digital twin integration.
Why this matters: The concept of an AI plane for 6G is not new but is gaining momentum as operators develop their AI-native strategies. e& UAE published an ISAC white paper in January and now follows with this architectural blueprint. The measurable AI-native KPIs are a useful contribution, moving the discussion from aspirational architecture to quantifiable targets. However, 80-page white papers from industry-academia partnerships need to be assessed against the concrete standards contributions they generate.
Implications for 6G standards: The paper explicitly references IMT-2030 and 3GPP Release 21 and beyond, and e& UAE stated it aims to shape global standards. The proposed AI-plane architecture aligns with the intelligence plane concepts in multiple 3GPP 6G Workshop submissions, including those from China Mobile and the FNS consortium.
View source material to read more
Selected research papers published this week
Multi-orbit space-based data centres for 6G
Shimaa Naser and colleagues propose a hierarchical architecture for 6G non-terrestrial networks in which LEO satellites handle radio access and real-time inference, while MEO and GEO layers provide regional aggregation, global orchestration and compute-aware routing. The framework embeds distributed in-orbit computing, energy-aware scheduling and AI-driven control across orbital tiers.
Why this matters: Direct handset-to-satellite communication is expected to be a core 6G capability, but managing massive device connectivity across dynamic LEO constellations creates coordination challenges that simple relay architectures cannot address. This paper moves the NTN discussion beyond connectivity into compute architecture, proposing that satellite constellations themselves become distributed computing platforms. The multi-orbit approach aligns with the terrestrial-satellite convergence that 3GPP is actively studying for Release 21.
View source material to read more
RIS deployment optimisation for urban 5G/6G networks
Adam Samorzewski and Adrian Kliks examine optimal placement of Reconfigurable Intelligent Surfaces (RIS) in the city of Poznań, Poland, to enhance radio signal quality in an urban 5G/6G network with eight base stations. Simulations cover both line-of-sight and non-line-of-sight conditions across a complex urban landscape.
Why this matters: RIS is one of the candidate technologies for improving coverage and capacity in future networks without deploying additional base stations. This study provides practical deployment guidelines for urban environments, demonstrating that strategic RIS placement can significantly reduce path loss and improve coverage in areas where conventional base stations alone are insufficient. The cost optimisation dimension is relevant to the economic sustainability concerns that operators consistently raise.
View source material to read more
Unified waveform framework for 6G physical layer
Xingyao Zhang and colleagues propose a unified multicarrier waveform framework for 6G, systematically categorising state-of-the-art waveforms into one-dimensional modulation types (such as OFDM and AFDM) and two-dimensional types (such as OTFS). The paper analyses their resilience against channel-induced interference and compares key performance indicators across diverse channel conditions.
Why this matters: Waveform selection is one of the foundational decisions for the 6G air interface. This framework provides a systematic basis for comparing candidate waveforms under different propagation conditions, which is directly relevant to ongoing 3GPP RAN1 discussions about 6G physical layer design. The inclusion of AFDM and OTFS alongside established OFDM reflects the broadening design space for 6G.
View source material to read more
Edge-enhanced healthcare coordination for 6G networks
Liuwang Kang and colleagues propose CAIP (Collective Adaptive Intelligence Plane), a standards-aligned coordination framework for mission-critical healthcare applications in 6G networks. The framework addresses workflow-level coordination under clinical timing constraints by mapping to existing 3GPP RRC, QoS/SDAP and O-RAN E2 interfaces, avoiding new protocol layers.
Why this matters: Healthcare is one of the most frequently cited 6G vertical use cases, but translating clinical timing requirements into network coordination abstractions remains largely unresolved. The backward-compatible approach, deployable on current 5G Advanced infrastructure, provides a practical migration path. The phased standardisation roadmap from xApp deployment to AI-native 6G specification is a structured contribution to the healthcare use case discussion.
View source material to read more
Multi-year spectrum measurements of 6G candidate bands
Amir Hossein Fahim Raouf and Ismail Guvenc present measurement-driven feasibility analysis of two US 6G candidate spectrum bands: 2.69–2.9 GHz and 4.4–4.94 GHz. Using software-defined radio measurements collected from 2022 to 2025, the study quantifies spectral contiguity, reliability and interference variability across multiple years.
Why this matters: These bands are central to US 6G spectrum policy, with the NTIA mandated to study them for potential commercial use. The finding that usable spectrum availability in the 2.7 GHz band dropped significantly between years (from near-unity to approximately 0.65) highlights the risk of assuming nominal bandwidth equals practical capacity. The 4.4 GHz band showed better structural stability with larger contiguous bandwidths, which is important data for WRC-27 preparations and US spectrum allocation decisions.