Before approving the next AI budget, check the network

AI has rapidly moved from experimentation to a boardroom priority.

But before approving the next AI budget, UK business leaders should ask a critical question: can their network support AI at scale once it moves beyond the pilot stage? Recent UK research found that 55% of organizations will prioritize AI or ML investment over the next 12 months, while nearly one in six admit they are investing aggressively with little evaluation because they fear being left behind.

This pressure is understandable, but this is where the ROI debate becomes too narrow.

The value of AI will not always appear neatly soon after deployment.

Some gains come through automation and productivity , while others emerge through long-term capability building: greater agility, better resilience, faster decision-making and new ways of working.

The danger is boards fund AI as a transformation, then judge it like a short-term software project.

In the UK, only 15% of organizations say their AI implementations have exceeded expectations, which should prompt a deeper discussion about the conditions AI needs to succeed.

Where AI performance breaks down When AI underperforms, the instinct is often to look at the model, the data, or the team that built the use case.

Those factors matter, but rarely explain the whole problem.

AI depends on the context it works in: the data, the network, the security model, the governance and the skills around it.

If those conditions are weak, even a promising use case can struggle to become durable and valuable in production.

A proof of concept can look convincing in a narrow environment, with clean data, controlled users and a clear route to value.

The difficulty starts when the same project has to operate across offices, clouds, systems and data rules that were never designed around AI.

This is when connectivity becomes part of the ROI calculation.