As European operators accelerate copper network switch-offs, replacing legacy infrastructure with fiber, a valuable opportunity is being overlooked. While the industry focuses on transitioning data, voice, and video services to fiber access networks—utilizing Point-to-Point fiber and XGS-PON technology—existing copper telephone pair and coaxial cabling within multi-dwelling units remains untouched and represents significant untapped potential. Rather than becoming obsolete, this in-building copper infrastructure can be repurposed to accelerate fiber adoption, enabling operators to serve homes more efficiently while allowing building owners to modernize properties and increase their value. Residents gain access to Gigabit broadband services without requiring complete rewiring. Historically, copper wiring was associated with legacy services including Plain Old Telephone Service and VDSL2 broadband. However, distinguishing between access network copper and in-building copper infrastructure is critical. As operators retire these services and fiber becomes the primary access technology, the existing wiring becomes newly available for innovative applications. This shift marks a transformative chapter where copper switch-off creates opportunities for leveraging existing infrastructure rather than simply signaling its technological obsolescence.

This paper presents a comprehensive survey examining the application of deep reinforcement learning (DRL) to 6G artificial intelligence radio access networks (AI-RAN). As 6G wireless networks progress toward deployment, the integration of advanced machine learning techniques becomes critical for optimizing network performance and resource allocation. The authors, including Jie Lu, Peihao Yan, Qijun Wang, Ruxin Lin, and Huacheng Zeng, explore how DRL algorithms can address complex optimization challenges inherent in next-generation wireless systems. The survey covers the theoretical foundations of deep reinforcement learning techniques and their practical implementation within AI-RAN frameworks, examining how these methods enable autonomous decision-making for spectrum management, interference mitigation, and network resource optimization. By synthesizing current research and identifying key technical approaches, the work demonstrates why DRL represents a pivotal technology for 6G networks, capable of adapting dynamically to changing network conditions and user demands. This research is significant for telecommunications professionals and researchers developing intelligent wireless infrastructure, providing essential insights into how machine learning will shape future network architectures and operational efficiency.

The transition from 5G to 6G represents a fundamental evolution in wireless communication, introducing ultra-high-speed connectivity, artificial intelligence-driven automation, and truly seamless global coverage. The 6G architecture integrates terahertz spectrum communications, reconfigurable intelligent surfaces, non-terrestrial networks, and quantum-secure communication protocols to enable transformative applications including holographic telepresence, digital twin networks, and autonomous systems. The network comprises five key layers: User Equipment, Access Network, Core Network, Edge and Cloud Computing, and Non-Terrestrial Networks, interconnected through standardized interfaces including Uu, Xn, N2, N3, N4, F1, N6, and NTN. These interfaces ensure seamless communication and ultra-low latency across network elements. The 6G system will leverage AI-powered core networks and advanced radio access network designs with terrestrial-satellite integration to deliver enhanced speed, reliability, and security. As 3GPP continues refining 6G standards, this intelligent and adaptive network architecture will enable unprecedented innovation in connectivity and support next-generation applications requiring extreme performance capabilities.

The telecommunications industry is undergoing significant restructuring as major operators pursue Open RAN and 5G technologies to escape vendor-locked Radio Access Network architectures. Open RAN, defined as a nonproprietary system enabling interoperation between equipment from different vendors, emerged primarily to counter Chinese vendor dominance and associated security concerns. Following the 2019 Bloomberg report of hidden backdoors in Huawei equipment deployed in Vodafone's Italian networks and subsequent Trump administration sanctions, Western nations prioritized network transparency. Open RAN promises cost efficiencies through standardized components, multi-vendor ecosystems, and reduced deployment expenses while fostering innovation. However, despite government subsidies and industry advocacy, commercial adoption remains surprisingly limited. The technology addresses fundamental industry challenges including vendor lock-in, inflexible customization, and high maintenance costs that have historically plagued centralized RAN models dominated by giants like Huawei and Ericsson. This examination traces Open RAN's origins, recent developments, and the persistent obstacles preventing widespread commercial implementation despite its theoretical advantages.

The Ukraine war exposed Europe's dangerous reliance on SpaceX's Starlink satellite constellation for critical military communications after Russian strikes devastated terrestrial infrastructure, revealing a broader vulnerability in dependence on privately owned global digital infrastructure. This dependency exemplifies how state security increasingly relies on technologies controlled by private companies whose commercial interests may diverge from government needs, a risk amplified by recent controversies surrounding access restrictions to AI models and NVIDIA's geopolitical positioning. Europe faces particular exposure given its dependence on external digital infrastructure, semiconductor companies, and telecommunications providers for defense and national security. While the European Union has attempted to address these deficiencies through initiatives like the 2023 European Chips Act, these efforts have proven insufficient, and tensions persist between government regulatory strategies prioritizing resilience and private sector commercial logic. To strengthen digital infrastructure resilience, Europe requires more than financial incentives. The solution demands binding coordination mechanisms between public and private entities, including conditionality on public funding, procurement leverage for demand-shaping, and mandatory rather than voluntary cooperation. The proposed European Tech Sovereignty Package and anticipated European Savings and Investment Union offer potential frameworks for implementing this more integrated approach to state-business alignment in critical digital infrastructure.

6G networks aim to transcend the limitations of current 5G infrastructure by integrating sub-terahertz frequency bands and native artificial intelligence to achieve sub-millisecond latency and terabit-per-second data rates. Unlike previous generations that relied on rigid, hardware-defined protocols, 6G shifts toward a software-centric model where the physical layer itself is optimized by deep learning algorithms to adapt to environmental interference in real-time.

The wireless communications industry increasingly relies on artificial intelligence and machine learning to optimize networks, enhance security, and manage spectrum, with these technologies proving essential for 5G advancement and future 6G development. However, the explosive growth of AI infrastructure has created unprecedented demand for memory chips—semiconductors that store rather than process data—used across connectivity and computing technologies. Manufacturers are strategically redirecting production capacity toward higher-margin memory products for AI data centers, creating a significant shortage of general-purpose memory chips essential for smartphones, computers, communications equipment, and vehicles. This shortage particularly affects Dynamic Random-Access Memory products like DDR4, which remain embedded in existing devices and network infrastructure despite manufacturers' focus on newer AI-oriented products. Unlike traditional supply-chain disruptions, this represents a deliberate reallocation of finite production capacity driven by strong economic incentives to prioritize advanced, high-bandwidth memory for AI customers over lower-margin consumer memory products. This development impacts the wireless sector at nearly every operational level, presenting a critical policy challenge: balancing rapid AI growth with maintaining resilient, affordable communications networks and accessible consumer devices.

Anvil Mobile provides comprehensive managed IoT connectivity services centered on multi-network SIM cards designed for machine-to-machine (M2M) deployments. Beyond basic SIM functionality, the company addresses critical infrastructure elements including network coverage, device monitoring, and customer support. IoT deployment enables real-time communication across industrial routers, smart sensors, cameras, and remote monitoring systems, automating processes that previously required manual intervention. This connectivity eliminates manual data collection, reduces response latency, and enables predictive maintenance and remote diagnostics, thereby reducing operational overhead. Anvil Mobile offers flexible SIM options including Static IP SIM cards for secure remote device access, Private APN configuration for isolated network paths, and SMS services for reliable device commands. The platform supports scalable deployments from single sites to thousands of endpoints without network re-architecture. The company distinguishes itself through consultative partnership, working with clients to understand their specific architecture and coverage requirements before deployment, then adapting connectivity solutions as fleet operations evolve. This approach ensures deployments are built on solid foundations rather than assumptions, with growth managed as a controlled event.

Three operators on three continents moved on AI Voice within six weeks. Ooredoo Qatar launched AI-powered outbound sales calling using Microsoft Azure OpenAI. Deutsche Telekom presented the Magenta AI Call Assistant at MWC Barcelona, built with ElevenLabs and embedded in the network rather than in an app. Orange Business followed with branded calling, deepfake detection and agentic telephony, while disclosing that it had already handled 80 million AI conversations across 7,000 enterprise customers in 2025. The products were different, but all three were making the same bet on Voice.

Every telco wants AI agents. Marketing wants one to reduce churn. Care wants one to resolve complaints. Finance wants one to optimize investment. The NOC wants one to find root causes. RAN wants one to optimize performance. There is only one problem. None of them necessarily agree on what a customer, service, network problem, or even revenue actually means. Giving them more data will not solve it. An LLM can read millions of records and an agent can call thousands of APIs, but neither inherently knows that Customer A owns Product B, which contains Service C, which depends on Resource D, which is currently affected by Alarm E, while Policy F prevents the obvious corrective action. Data contains the values. Ontology gives those values meaning, relationships, and constraints.

It is worth remembering our glorious past. Bell Labs invented the transistor in 1947 because vacuum tubes were becoming a maintenance problem for the telephone network, and in trying to fix telephony it accidentally launched the semiconductor industry. Claude Shannon wrote information theory there. Its scientists made foundational contributions to the laser, built Telstar for its 1962 launch, developed the CCD image sensor, and wrote Unix and the C programming language.

Si les États-Unis demeurent la principale puissance scientifique, leur position est fortement contestée par la Chine, particulièrement en Sciences Physiques et de l’Ingénieur où cette dernière occupe nettement la première place. La part d’audience de la France se détériore, passant de 4,3% à 2,7%, soit une baisse de 37,2% sur les deux dernières décennies. Cette baisse est la plus importante du panel après celle du Japon. En cinquième position en 2005, la France occupe aujourd’hui la dixième place, ayant été progressivement devancée par plusieurs pays comparables tels que le Canada, l’Italie et l’Australie, notamment en période post COVID-19. Elle est désormais rattrapée par des pays tels que l’Espagne, la Corée du Sud ou les Pays-Bas, dont l’audience pesait moitié moins que la sienne il y a vingt ans.

The story begins with a technology called Meshtastic. You can download Meshtastic firmware and flash it to a Long Range (LoRa) mesh device. LoRa devices are inexpensive and readily available through multiple sources, including Amazon. These devices typically run RF output power of 22 dBm or about 158 milliwatts. The Meshtastic device connects to your mobile phone using Bluetooth along with an app you download, which allows you to send and receive text messages over the mesh network. If you have a few of these devices, you can communicate with each other over short distances. They’re handy to have on a group hike in a park or on a mountain where there’s spotty or no cellular service, for example.

Like it or not, you are already distributing billions of AI tokens every day. To enterprises, to consumers, and increasingly to your best new customers, which are not people at all. AI agents are already on your network. Humanoids, delivery robots, drones, and every other intelligent machine anyone can imagine will follow, and they will all move their atomic units of intelligence across your infrastructure.

I am working with a Tier One telecom operator on its agentic AI strategy. Much of the work has nothing to do with choosing a model or buying more GPUs. We are trying to understand what the customer becomes when software starts making decisions, buying services, using networks, and acting with authority on somebody else’s behalf. Telecom has spent decades designing around humans holding phones and machines sending traffic. The next customer may be an agent that sits between both. Several pieces are already real. Agents can operate computers, retain memory, write software, use payment rails, and run across distributed inference infrastructure. Those are no longer research demos dressed up as strategy slides. They are shipping capabilities.

Telco security teams are used to ridiculous cyber numbers. For example, in the first six months of 2026, Cloudflare blocked 23.2 million network DDoS attacks across its infrastructure. That is about 128,000 attacks every day, or more than 5,300 every hour. It also recorded 935 attacks above one terabit per second during the same period. The volume is so high, and many attacks are so short, that human response became unrealistic years ago.

The article argues that current AI policy discourse suffers from narrow focus on visible infrastructure layers—chips, cloud, and models—while overlooking critical strategic dynamics lower in the technology stack. China is pursuing a fundamentally different AI strategy centered on deployment rather than frontier model development, leveraging US export controls to position itself as the default infrastructure supplier for physical AI systems across the Global South. Through dominance in wireless networking, device manufacturing, and critical raw materials like magnets and batteries, China threatens to create upstream dependencies for Western markets while excluding foreign competitors from its own. However, the article identifies vulnerability in China's model: its value creation depends on global market access. A trusted technology coalition among allied nations could exploit this dependency by establishing secure network standards and creating ex-China markets for key technology segments. The emergence of hyperconnected 6G systems, with integrated sensors and edge devices connected to critical infrastructure, creates shared cybersecurity risks that provide strategic justification for such coalition-building. The article concludes that only through coordinated alliance action can the US effectively counter China's asymmetric positioning and secure technological sovereignty.

AT&T has officially commenced phasing out its copper POTS (Plain Old Telephone Service) lines, infrastructure that has supported critical systems for over a century. The FCC accelerated this transition in March 2026 by removing regulatory barriers, including requirements for network change disclosures and case-by-case filings, enabling carriers to discontinue legacy services more rapidly. AT&T's shutdown timeline spans 2025 to 2029, with the company stopping acceptance of new orders for copper-based services across nearly 20 states as of October 2025. A recent FCC filing reveals AT&T seeks authority to discontinue legacy TDM-based voice services affecting approximately 90,000 customers across 18 states by November 2026. Organizations relying on POTS lines for fire panels, elevator phones, and alarm systems must urgently identify analog dependencies and develop replacement strategies before regional retirements create disruptions. AT&T expects to retire copper-based services across the majority of its footprint by the end of 2029, as confirmed by CEO John Stankey and Executive Vice President Susan Johnson.

While 5G networks continue their global rollout, researchers are advancing the technical foundations for sixth-generation wireless communication (6G), anticipated for commercialization around 2030. Rather than a mere speed enhancement, 6G represents a paradigm shift from connecting devices to connecting intelligence. The technology targets terabits-per-second data rates, sub-millisecond latency, and ultra-reliable communication to enable real-time holographic communication, multi-sensory digital twins, and ubiquitous AI applications. Key technical enablers include terahertz band frequencies (0.1 to 10 THz) for massive bandwidth, though these require novel transceivers and channel coding solutions due to atmospheric absorption challenges. Artificial intelligence will replace traditional signal processing, dynamically adapting the physical layer to channel conditions. Reconfigurable intelligent surfaces (RIS) will transform buildings into smart reflectors using programmable metamaterials. Integrated sensing and communication capabilities will provide centimeter-level precision detection for autonomous systems. The distributed core network architecture will replace traditional cell towers with cooperative access points ensuring seamless coverage. Post-quantum cryptography addresses security threats from quantum computing advances, positioning 6G as a foundational infrastructure for next-generation intelligent systems.

Telefónica has released the final installment of its Digital Networks Act analysis, outlining significant concerns about the proposed regulation's approach to telecommunications infrastructure resilience. While the EU and telecommunications sector unanimously recognize the strategic importance of securing digital networks essential to economic and public services, the DNA's resilience provisions create problematic regulatory duplication. Articles 4 to 8 of the DNA impose sector-specific requirements that substantially overlap with obligations already established under the NIS2 Directive and the Critical Entities Resilience Directive, introducing complexity and legal uncertainty rather than simplification. Additionally, the proposed Single Passport system for European authorization could paradoxically complicate governance by enabling jurisdictional conflicts when different national authorities oversee cybersecurity matters for operators authorized elsewhere. The proposal further fragments oversight by involving multiple actors—including National Regulatory Authorities, BEREC, the future Office for Digital Networks, and national cybersecurity authorities—potentially creating inconsistent compliance requirements and reporting obligations. Rather than strengthening the regulatory framework, these overlapping mechanisms risk undermining it.