Future telecoms 2025 to 2026: from faster networks to thinking infrastructure

By Marc Ambasna-Jones 19 Dec 25 Reading time: 7 mins

When Professor Dimitra Simeonidou founded the Smart Internet Lab in Bristol in 2015, her big concern was industry fragmentation. Optical, wireless, cloud and application communities were all advancing rapidly, she argued, but almost no one was looking at how the whole system fitted together end to end.

“Networks were getting more complex, more software-defined, and more critical to society, yet our research and infrastructure were still organised in silos,” she told BI Foresight at the Lab’s 10th birthday in November.

Today that observation feels increasingly well-judged, because 2025 exposed just how interdependent modern networks have become. Efforts to shape 6G have moved beyond abstract roadmaps towards practical questions about energy use, automation, and resilience. Platforms such as JOINER have brought together radio, fibre, compute, and applications in shared test environments, while live demonstrations, including a quantum-secured video link between Bristol and Cambridge [see boxout, below] showed how new security techniques can operate on real networks rather than in isolation.

At the same time, operators and vendors have faced growing pressure to cut energy consumption even as data volumes have risen. Progress in one part of the network now depends on how well the whole system is designed and operated. The pieces needed to support future technologies such as 6G and quantum communications are beginning to come together, but only if telecoms is treated as an integrated system rather than a collection of parts.

AI moves into the fabric of the network

By 2025, artificial intelligence was no longer confined to analytics dashboards or post-hoc optimisation. Operators and vendors were increasingly using software to make live decisions about how networks were run, how capacity was allocated, how faults were detected, and how performance was balanced against energy use. Modern networks are now too large and too dynamic to be managed manually, so AI has become a logical next step in helping to manage multiple network functions.

What became clearer this year, however, is that adding intelligence does not automatically simplify things. AI systems trained to optimise different parts of the network can work at cross-purposes, improving performance in one area while increasing cost or energy consumption in another. The result is not necessarily failure, but almost certainly a growing recognition that automation only works when it is designed end to end, rather than bolted onto existing silos.

This idea has shifted the conversation, somewhat, from how quickly AI can be deployed to how it should be governed. Questions of data quality, accountability, and energy impact moved closer to the centre of network planning in 2025, particularly as operators looked ahead to more autonomous networks and the longer-term demands of 6G.

But this shift has exposed a hard truth that intelligence consumes energy and that has an impact. Operators are discovering that deploying AI indiscriminately can erode the efficiency gains it promises. Multiple models running in parallel, each responding to local conditions, risk increasing power consumption and operational complexity rather than reducing it.

As a result, 2025 saw a more pragmatic conversation emerge around de-risking AI, not only in terms of safety and governance, but also energy use, reliability, and accountability. The challenge for 2026 will be deciding where AI genuinely earns its place in the network, and where simpler approaches may be more effective.

6G: from generational upgrade to infrastructure rethink

These pressures have fed directly into how the industry is approaching 6G. In 2025, discussion of the next mobile generation became noticeably more restrained. Rather than focusing on headline speeds or consumer use cases, attention shifted towards what future networks would actually need to deliver, and what constraints they would have to operate within.

At events such as the Brooklyn 6G Summit, researchers and vendors were explicit that 6G is not being designed as a faster version of 5G. Peter Vetter, head of core research at Nokia Bell Labs, summed up the change in emphasis.

“Every generation, capacity goes up by a factor of 10, and the new technology for 6G is AI.”

In Vetter’s view, the centre of gravity moves away from devices and towards the network itself. Intelligence, autonomy, and coordination become defining features, shaping how resources are allocated, how services are prioritised and how networks respond to changing conditions. But there are also challenges. As Vetter adds, “you want capacity to go up by 10 times, but not the power consumption.”

This energy challenge has become central to 6G thinking. Mobile networks already account for the bulk of telecoms energy use, and few operators see a path to deploying a new generation that increases operating costs or undermines net-zero commitments. As a result, 2025 saw growing emphasis on architectural efficiency, from more selective use of spectrum and smarter antenna systems to tighter coordination between radio, fibre, and compute.

Another strand gaining traction is integrated sensing, the idea that network infrastructure can also observe aspects of the physical environment. Bell Labs has demonstrated the ability to localise people with metre-level accuracy and detect drones beyond visual range, using network signals rather than dedicated sensors. While still early, this points towards networks that support safety, transport, and industrial systems alongside traditional communications.

What links these developments is a shift in how 6G is being approached. It is no longer treated simply as a new radio technology, but as an overhaul of how networks are built and run. Its success will depend on whether intelligence, energy use, security, and automation can be made to work together, rather than being optimised in isolation. That brings the industry back to the concern Professor Simeonidou raised a decade ago that without an end-to-end view, advances in one area quickly create problems in another.

Quantum edges into the frame

Alongside AI and energy, quantum technologies moved closer to the mainstream of telecoms discussion in 2025. Field trials of quantum networking over commercial fibre, progress in quantum-secure communication, and interest in hybrid classical-quantum architectures all suggest a growing role for telecoms infrastructure in future security and resilience.

In the run-up to the Connected Futures Festival earlier in 2025, a quantum-secured video call was successfully carried out between Bristol and Cambridge, linking ProfessorSimeonidou in Bristol with Professor Richard Penty in Cambridge.

The call demonstrated how quantum security techniques can be integrated into live telecoms infrastructure rather than confined to laboratory settings. It combined optical networking, quantum key distribution and conventional communications systems to protect a real-time video link.

The demonstration was later shared and discussed at the Connected Futures Festival, where it served as a practical example of how quantum technologies might be deployed incrementally, alongside existing networks, for sensitive and critical communications.

Looking to 2026: AI, quantum, and shifting perspectives

Across operators, vendors, and policymakers, there is a growing sense that the industry is moving beyond digital transformation into something more demanding. AI is being embedded into day-to-day operations. The question for 2026 is where AI will genuinely change networks for the better, and where it simply adds cost, or risk.

What follows from that is not a rush towards more technology, but a more selective approach to innovation. In 2026, progress is likely to be driven less by standalone breakthroughs and more by how well different technologies are made to work together. AI, automation, security, and energy efficiency are increasingly being developed as connected concerns rather than separate lines of research and development.

This is already reshaping where innovation effort is directed. Rather than chasing general-purpose intelligence in the network, much of the focus is moving towards narrow, context-specific applications, such as automating fault resolution, managing energy use, coordinating traffic across radio and fibre, or supporting specialised enterprise services. The emphasis is on systems that behave predictably under pressure, rather than those that simply demonstrate technical sophistication.

Convergence will therefore remain a defining feature of innovation in 2026. Networks are being asked to support AI-driven services, hybrid classical–quantum security, satellite integration, and new sensing capabilities at the same time. Innovation increasingly sits at the boundaries between these domains, where design choices in one area directly affect performance and resilience in another.

In 2026, the most valuable advances are likely to be those that reduce friction – between technologies, between organisations, and between competing demands on the network – rather than those that add yet another layer of complexity.

Marc Ambasna-Jones / Editor

Working as a technology journalist and writer since 1989, Marc has written for a wide range of titles on technology, business, education, politics and sustainability, with work appearing in The Guardian, The Register, New Statesman, Computer Weekly and many more.