The week in AI-native telco (19-25 Sept 2026)

• Voice was at the heart of AI-native telco developments during the week spanning 19-25 September 2026
• AI infrastructure energy consumption, AI-RAN, sovereign AI and wearables were also hot topics

There’s so much going on in the AI-native telco world that we’ve decided to package all the main developments up into a weekly roundup that sports the highly imaginative title of The Week in AI-Native Telco – or, if you like, WANT. 

So, what were the main developments from the seven days that ended on Friday 25 September? Here’s the pick…

In the latest example of a telco developing new revenue-generating, AI-enabled, network system-based applications, Vodafone Group and vendor partner Ericsson have successfully demonstrated two new customer-centric voice services on a test mobile network. The first enables instant language translation during standard voice calls, without customers requiring an app or modified smartphone. Vodafone noted that engineers from the partner companies “were able to instantly translate conversations between people speaking in English, Spanish, Italian, French and German,” and that the service could “be easily expanded to support additional languages.” The second is real-time noise cancellation, a service that “intelligently identifies and removes background noise in real time, ensuring call clarity for users in loud environments like busy streets, public transport or construction sites”. Read the full article.

Liberty Global has struck a three-year strategic partnership with Sierra to support the rollout of the conversational AI platform developer’s technology across Liberty Global’s operations in Europe (Belgium, Ireland, the Netherlands and the UK) which, combined, have approximately 80 million fixed and mobile customer connections. The telco group says the agreement will provide its businesses “with a common approach to deploy AI agents capable of engaging customers in natural language, across chat, voice and text,” adding that it is “accelerating the use of AI across its businesses to make customer interactions easier, more intuitive and more effective.” It added: “By enabling AI agents to handle routine and increasingly complex interactions, the partnership will also allow customer-care teams to focus more of their time on situations requiring human judgement and deep expertise.” 

Liberty Global’s UK operation, Virgin Media O2 (VMO2), has made its own announcement about how it will use Sierra’s technology. The UK operator, which is jointly owned by Liberty Global and Telefónica, has unveiled a new AI voice agent that will support calls from customers experiencing faults with their broadband service. The aim, according to the UK telco, is to help customers get faster help with routine issues, while freeing up human agents to focus on more complex cases. Customers who want to speak to a human agent will still have that option, noted VMO2, and all calls will be monitored by its team. It is part of a wider customer service transformation programme at VMO2 that recently saw the operator launch a specialist 500+ strong UK-based team to deal with the most complex customer queries. It has also partnered with the Institute of Customer Service on a structured and officially recognised training programme designed to equip contact centre employees with the skills to better manage complex calls. VMO2 has also invested in AI tools for human agents, including analytics tool Lumi AI, which provides them with real-time prompts and insights during a customer call.

Here’s an important development for all the companies, including telcos, that are building and running AI infrastructure bigger than 500 kilowatts (KW) (about 230 square metres) in European Union countries… The European Commission has proposed a common rating scheme for datacentres “aimed at increasing transparency on their energy use and supporting their sustainable integration into Europe’s energy system”. The EC noted that “datacentres are essential to Europe’s digital future” as “increasing computing capacity in Europe is important for strengthening Europe’s technological independence and digital sovereignty,” which is “why the EU wants to triple its datacentre capacity over the next five to seven years”. However, such growth “brings challenges which need to be addressed. In addition to using more energy, an increase of datacentres could put extra pressure on electricity grids and the use of natural resources, such as water, and contribute to increased carbon emissions. To limit these impacts, it is essential that datacentres are highly energy- and water-efficient and sustainably integrated in the energy system.” To help achieve those goals, the EC’s proposed rating scheme “will help ensure that there is transparent information about the actual use of resources (energy, water) by datacentres. The rating scheme will cover individual datacentres with a capacity above 500 KW. It will also cover datacentres’ contribution to the grid, such as the reuse of waste heat, the addition of clean energy-generation capacities as well as flexibility.” The EC notes that reusing around half of all waste heat from European datacentres is “the equivalent of the total heating demand of 4 million households in Europe. Sustainable, flexible and well-designed datacentres that can adjust their electricity use to the grid’s conditions, can help lower the overall cost of the electricity system, improve grid stability, and integrate more renewable energy.” The EC has now launched a call for evidence and public consultation on minimum performance standards for datacentres operating in EU markets.

MeetKai, a “sovereign AI” company that builds and operates national AI platforms for governments and telcos, has announced a major rollout of its sovereign AI stack, which is built on Nvidia technology, in six countries, and is teaming up with Veon in five of those markets. In Brazil, MeetKai is working with IT services company Serpro, while in Ukraine, Pakistan, Kazakhstan, Uzbekistan and Bangladesh it is working with Veon. Across Veon’s five markets, home to more than 150 million mobile customers and over 225 million digital customers, MeetKai will act as Veon’s “AI lab-as-a-service”, beginning with Kyivstar in Ukraine, followed by Pakistan, Kazakhstan, Uzbekistan and Bangladesh, with each site operated by MeetKai together with the local telco partner. James Kaplan, co-founder and CEO of MeetKai, stated: “Sovereign AI only works when the entire AI stack, infrastructure to models, sits inside the country. That is a hard thing to build once and a very hard thing to build six times. Building on Nvidia turns it into a repeatable design: The same accelerated computing under every site, our platform and national models on top. Ukraine will have AI it fully controls, Brazil takes the same approach into Latin America on the back of our sovereign AI win there, and we will replicate it in every market in between.”

Samsung Electronics has signed contracts with South Korean telcos KT Corp. and SK Telecom for their private network AI-RAN projects under the Korean Ministry of Science and ICT (MSIT)’s Hyper AI Network initiative that brings together network operators, technology vendors and other industry partners. Samsung will be the sole global vendor for KT Corp, and the main vendor for SK Telecom. “Building on its decades-long partnerships with the country’s leading operators and its leadership in software-driven, AI-native network innovation, Samsung is the only partner selected for both projects,” the company noted. Set to begin in October 2026, the projects will “deploy 5G Standalone (SA) private networks across industrial environments, ranging from shipyards to petrochemical facilities, to validate AI-RAN capabilities and generate real-world insights for future deployments,” noted Samsung. Read the full article.

With telcos such as Verizon not only keeping an eye on the impact that AI wearable devices will have on their network data traffic flows but also teaming with emerging vendors such as Snap, it’s interesting to note that sales of AI glasses are ramping up at a furious pace. Research firm Omdia notes in this press release that global shipments of AI glasses reached 4.2 million units during the first half of 2026, an increase of 127% compared with the same period a year earlier. Meta is the clear market leader, shipping 3.4 million devices for an 81.3% market share in the first half of 2026, though smaller competitors, such as Shenzhen-based Even Realities (2.4% market share) and other Chinese vendors, including Alibaba (2.1% market share), are starting to have an impact. Qiran Ju, senior analyst at Omdia, noted: “Growth in the AI glasses market in [first half 2026] was driven by Meta’s continued market momentum and a wave of new market entrants from China. Just as critically, vendors have made the value proposition immediately clear to buyers by focusing on hands-free photo and video capture, open-ear audio, and AI assistance within everyday frames.” The analyst added: “Even Realities takes a different frame design from most AI glasses. By removing both the camera and speakers and focusing instead on a slimmer design, a small monochrome display and productivity features, such as real-time transcription, it shows how vendors are exploring different ways to make AI glasses useful in everyday life.”

T-Mobile US is expanding its deployment of Autopilot, an “intent-based AI automation” application, as part of its self-organising network (SON) platform to enable its 5G network to automatically make adjustments as conditions change, for example during storms. “Recent testing showed AutoPilot can make real-time network adjustments in about half the time, helping the network respond faster when conditions change,” noted T-Mobile US in this announcement. The operator is also rolling out Dynamic CX, which “uses AI to anticipate demand ahead of major events and automatically optimise network performance”. Both Autopilot and Dynamic CX are now being deployed nationwide. John Saw, CTO at T-Mobile US, noted: “The next era of network resilience is about building a network that can think, adapt and act faster in the moments that matter. With AutoPilot, we’re taking the intelligence already built into our self-organising network and giving it even more ability to respond as conditions change, whether that means helping close a coverage gap caused by extreme weather, optimising performance or keeping customers connected during unexpected disruptions. That’s what makes a great 5G network stronger: Not just more capacity but more intelligence behind every connection.” The broader deployments are part of T-Mobile US’s preparations for the anticipated disruption caused during the upcoming storm season in the US, with the operator also noting it is making ongoing investments in backup power and diverse network connectivity options in order to help keep customers connected for as long as possible.

Telefónica has unveiled a new AI-enabled observability service aimed at helping its business customers gain real-time insights into the performance of their networks. The new offering, which is aimed at larger enterprises and public sector organisations, uses AI capabilities to unify network monitoring, analysis and event correlation into one platform. The Spanish telco said the new solution takes capabilities that were previously available as part of its Titan Connect advanced connectivity solutions portfolio to support organisations in recording the impact that network performance has on their business operations, and to anticipate potential incidents through data-driven insights. Eduardo Ariste, CEO of B2B and public sector at Telefónica España, said: “With this approach, we provide better service to the companies that work with us, enabling them to move from reactive to proactive management aligned with their business objectives, while reinforcing our position as the best gateway for enterprises to access digital technologies.” The service includes a dashboard designed as an operational management tool for IT teams, which combines data from LAN, WAN and Wi-Fi observability. It is being offered in two versions: The essential package which enables solution customisation based on operational requirements, or the premium package, which will be available from December and comes with all the abilities included in the essential package, plus open telemetry support, a data access application programming interface (API), an IT service management API that automatically generates incident tickets, and customisable reports.

SK Telecom has developed digital twin technology in collaboration with the Korea Institute of Science and Technology Information (KISTI) to verify the performance and operational feasibility of a quantum cryptographic communication network prior to the network being built. SK Telecom noted in this announcement (in Korean) that once a quantum cryptography network is installed, it is difficult to adjust the deployed elements, such as optical cable length, or change parameters, such as optical loss. In addition, the equipment used for such deployments is expensive, so it is uneconomic to test various setups. But, according to SK Telecom, the digital twin it has created with KISTI has “virtually replicated this entire testing process”, enabling network operators to identify the optimum architecture and settings for the required quantum cryptographic communication network.

Verizon has announced a further $50m of funding for an upskilling initiative that aims to help workers and job seekers across America improve their AI skills. The Verizon AI Skills for America scheme builds on the $20m of investment the US carrier has already made available for its Reskilling and Career Transition Fund for departing workers.  Combining this with the new funding to provide access to best-in-class AI training takes the total investment in the initiative to $70m so far. Verizon will launch what it calls a “one-stop-shop” portal that brings together training content from AI experts, such as IBM, Google, Microsoft, Coursera and OpenAI, at no cost to participants. The operator will also work with trusted mission-based organisations, such as the Local Initiatives Support Corporation, the National Association for Community College Entrepreneurship (NACCE), and Goodwill Industries International, to offer hands-on coaching and practical AI guidance. “Creating meaningful economic opportunity in the AI era is not something any one company can achieve alone,” said Donna Epps, chief responsible business officer at Verizon. “We are deeply grateful to our tech partners for collaborating with us to bring world-class AI training to local communities entirely free of charge.” 

- The staff, TelecomTV