What AI thinks telecoms will look like in 18 months

Predictions about the next transformative technology in telecoms arrive with predictable regularity. Whether the topic is 3G, 5G, 6G, network APIs, satellite connectivity or autonomous networks, industry conferences and keynotes rarely lack for speculation about what comes next. Rather than add another forecast to the pile, we asked artificial intelligence itself to assess what the sector might resemble 18 months from now—and where operators could stumble.

The programmable network emerges

When asked what will feel different for the telecoms industry by early 2028, AI offered this perspective: "The biggest change may be that the network stops being treated primarily as connectivity infrastructure and starts being treated as a programmable platform."

According to the AI analysis, operators will increasingly expose network capabilities through APIs, allowing dynamic adaptation of connectivity to application requirements and leveraging AI to handle expanding portions of network operations. "The customer may not notice any of this directly, that's the point."

Rather than requesting generic "5G" connectivity, enterprises will seek specific outcomes: guaranteed latency, additional uplink capacity, precise positioning, temporary coverage or defined reliability levels. "The network becomes something applications can interact with rather than simply something they connect to."

AI for networks versus networks for AI

A critical distinction exists between two applications of artificial intelligence in telecoms. AI for networks involves deploying AI to operate telecom infrastructure with greater sophistication: anticipating failures, fine-tuning radio resources, cutting energy use, enhancing customer support and streamlining network operations.

Networks for AI represents the more disruptive opportunity. "AI applications increasingly need connectivity with predictable latency, high uplink capacity, reliability and access to network capabilities such as location and authentication. That means the network itself becomes part of the AI application's architecture."

Autonomy with guardrails

When pressed on whether AI will eventually run telecoms networks, the AI cautioned against oversimplification. "Eventually, parts of it may. But, 'AI will run the network' is one of those statements that sounds more dramatic than it is useful."

Telecoms networks operate in unforgiving environments where mistakes carry severe consequences. Unlike recommendation systems that can recover from poor suggestions, network errors can disconnect thousands or millions of users. "The near-term future is therefore more likely to be AI-assisted autonomy rather than unrestricted autonomy. AI will increasingly recommend, predict, optimise and execute within defined boundaries."

The genuine challenge will not involve making AI capable of action. Rather, "The difficult part won't be making AI capable of taking actions. It will be deciding which actions operators are comfortable allowing it to take."

The evolving role of network engineers

Automation will transform rather than eliminate network engineering roles. "Network engineering is unlikely to disappear, but the balance between manual intervention and system-level orchestration will shift."

Instead of manually diagnosing unusual behaviour in individual network components, engineers will increasingly direct AI systems to investigate problems across multiple vendors, domains and datasets. "The human role moves towards supervision, judgement, architecture and exception handling."

This shift could elevate telecoms engineering to a more strategic function. However, a significant risk looms: "If organisations automate the routine work without investing in the skills needed to supervise increasingly autonomous systems, they could create a new operational vulnerability: nobody fully understands the system anymore."

5G's persistence and 6G's trajectory

Despite industry enthusiasm for 6G, 5G will remain a central topic through 2028. "5G is not disappearing just because 6G is being developed. In fact, the important question is whether operators complete the transition towards the capabilities of 5G Standalone and 5G-Advanced that make the network more programmable and suitable for advanced applications."

6G represents more than incremental speed improvements. "It is more likely to represent a shift towards networks designed from the outset around AI, programmability, sensing, automation and new forms of connectivity."

Satellites as complement, not replacement

Satellite connectivity will not supplant terrestrial mobile networks but rather complement them. "The more interesting future is one where customers don't need to know which network they're using. Your phone might move between terrestrial mobile, Wi-Fi and satellite connectivity according to availability and requirements."

In emergencies, rural regions and remote locations, satellite infrastructure could operate as effectively invisible backbone support.

The industry's blind spot

When asked what the telecoms sector might misunderstand about its own trajectory, AI offered this observation: "That the future of telecoms is not really about networks. It's about what becomes possible because networks can understand, predict and respond to what is happening around them."

The next phase will fuse connectivity with AI, compute, sensing, identity, security and automation. "The network becomes an active participant rather than a passive pipe."

Yet a warning accompanies this vision: "Telecoms should not become so obsessed with making networks intelligent that it forgets to make them useful. The winning network won't necessarily be the one with the most advanced technology. It will be the one that makes something valuable possible."

Beyond the technology

Substantial advances loom across AI capabilities, 6G development and satellite connectivity. Yet technology alone tells only part of the story. For operators and the broader sector, converting these developments into tangible impact presents the genuine test—whether through cost reduction, new revenue channels, improved customer experience or network enhancement.

The coming 18 months may not revolve around a single technology achieving dominance. Rather, multiple innovations working in concert could drive smarter and more efficient networks, though none will serve as a universal solution.