Why AI’s Next Challenge Looks Surprisingly Familiar to Telecom

Over the past year, one shift has become increasingly apparent in conversations around enterprise AI. The discussion is no longer dominated by which model performs best or where AI can automate another process. Those questions still matter, but they are no longer the deciding factor in whether AI moves beyond a pilot. Increasingly, organisations are asking a different question: what has to change inside the business before AI can be trusted with greater responsibility?

It’s a subtle shift, but one that says a great deal about where the market is heading.

For much of the past two years, the AI industry has focused on improving models. Accuracy has increased, reasoning has become more sophisticated and every new release has expanded expectations about what AI can achieve. At the same time, enterprises have been experimenting with how those capabilities can improve customer experiences, streamline operations or uncover new revenue opportunities. McKinsey’s latest global survey reflects just how quickly AI has entered mainstream business. Nearly nine in ten organisations now use AI in at least one business function, yet relatively few have successfully scaled AI across the enterprise. The technology is progressing rapidly. Building organisations that can use it responsibly is proving to be a different challenge altogether.

The Shift from AI Capability to Operational Execution

The questions now emerging have less to do with intelligence and much more to do with execution.

What happens when AI doesn’t simply recommend the next best action but initiates it? How should commercial policies be applied when decisions are made in real time? Who remains accountable if an autonomous system makes the wrong commercial judgement? How do organisations demonstrate that customer permissions have been respected, regulatory obligations have been met and every decision can still be explained months after it was made?

These questions aren’t appearing because AI has stopped improving. They’re appearing because businesses are becoming more confident about placing AI closer to the centre of their operations.

An AI model can identify the right offer for a customer, recommend a financial product, optimise network resources or determine the next step in a digital journey within seconds. Turning that recommendation into a business outcome is considerably more complex. Before anything happens, the organisation still needs to establish whether the customer is eligible, whether consent has been captured, whether commercial policies allow the transaction to proceed, whether another ecosystem partner is involved and whether every stage of the process remains transparent enough to satisfy governance and regulatory requirements.

That operational layer rarely attracts the same attention as the model itself, yet it increasingly determines whether AI can be deployed confidently at scale. Deloitte’s 2026 State of AI research found that almost three-quarters of organisations expect to adopt agentic AI within the next two years, while only a small minority believe their governance capabilities are mature enough to support autonomous AI at enterprise scale. The gap isn’t one of innovation; it’s one of operational readiness.

Looking across enterprise deployments today, a consistent pattern is emerging. Organisations aren’t struggling because AI models fail to generate recommendations. They’re discovering that every recommendation has to pass through the operational reality of the business before it becomes an action. AI doesn’t replace that environment. It becomes another participant within it.

Why Telecom Infrastructure Holds the Key to AI Governance

For communications service providers, that challenge is remarkably familiar.

Telecom has spent decades operating digital businesses where trust is established continuously rather than assumed. Networks verify identities, authenticate devices and provide trusted signals in real time. Alongside them, digital business platforms govern customer relationships, apply commercial policies, manage digital entitlements, orchestrate services, coordinate partner ecosystems and ensure transactions can be fulfilled, charged and settled correctly. Those capabilities weren’t designed with generative AI in mind. They were developed because digital businesses have always depended on trusted interactions between customers, networks, enterprises and partners.

As AI becomes embedded within those same environments, the value of that operational foundation becomes much more visible.

One assumption that deserves challenging is the idea that safer models automatically lead to greater enterprise adoption. Experience suggests something rather different. The closer AI moves towards business-critical decisions, the more organisations scrutinise everything surrounding the model. Questions about identity, governance, customer permissions, commercial accountability and policy enforcement become significantly more important because the consequences of getting those decisions wrong become significantly greater.

This is where telecom has a distinctive contribution to make.

For years, communications providers have been refining the operational capabilities required to establish trust across complex digital ecosystems. Networks contribute trusted identity and real-time intelligence. AI-native BSS and OSS platforms provide customer context, commercial governance, service orchestration, digital commerce and lifecycle management. Together, they create the operational environment that determines whether an AI-generated recommendation should become a trusted business action.

The same shift was reflected during Tecnotree’s discussion at Mobile World Congress earlier this year on autonomous networks. Interestingly, the debate wasn’t centred on whether autonomous systems would eventually make decisions independently. That increasingly feels inevitable. The discussion focused on governance, accountability and commercial oversight once those decisions begin happening faster than people can realistically intervene. Those are no longer telecom-only questions. They’re becoming enterprise AI questions.

Bridging the Gap Between AI Intelligence and Trusted Business Outcomes

This broader perspective also explains why initiatives such as GSMA Open Gateway have gained momentum. Standardised APIs exposing capabilities such as Number Verification, SIM Swap and Device Location provide trusted network intelligence that strengthens digital interactions. Their greatest value, however, isn’t realised through network APIs alone. It emerges when trusted network capabilities are combined with digital business platforms capable of understanding customer context, applying commercial policies, orchestrating services across multiple partners and governing the entire customer lifecycle. AI may determine the next best action, but the surrounding business environment ultimately determines whether that action should happen at all.

Viewed this way, the Trust Economy becomes far more practical than a discussion about secure AI. Trust isn’t created because a model generates a convincing recommendation. It is reinforced continuously as identity is verified, customer permissions are respected, policies are enforced, commercial obligations are met and every participant in a digital interaction remains accountable for the outcome. The model contributes intelligence. The surrounding ecosystem provides the confidence to act on it.

For communications service providers, this represents an opportunity that extends beyond deploying AI inside their own operations. The industry has spent years investing in trusted digital infrastructure, AI-native business platforms and ecosystem capabilities designed to support increasingly sophisticated digital interactions. As enterprises move from experimenting with AI to operationalising it, those capabilities become relevant in ways few anticipated. Gartner expects that by 2027, many organisations will scale back or decommission autonomous AI initiatives because governance and operational controls fail to mature at the same pace as the technology itself. That should be viewed less as a warning about AI and more as a reminder that successful adoption depends on much more than model capability.

The next chapter of AI will undoubtedly bring more capable models, but capability alone won’t determine which organisations succeed. The defining advantage will come from creating digital environments where intelligence, governance, trusted networks and business platforms work together seamlessly. Organisations will increasingly judge AI not simply by the quality of the answers it generates, but by whether those answers can become trusted outcomes.

Telecom understands that challenge because it has been solving comparable operational problems for decades. As the Trust Economy continues to take shape, the combination of trusted networks, AI-native BSS and OSS platforms, and digitally orchestrated ecosystems places communications providers in a unique position to help enterprises move AI from experimentation to responsible execution.