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.
CIO, Expereo.
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. If data moves too slowly, workflows break between systems, response times vary by location, or teams cannot see where performance is degrading, the business starts paying for delays, rework and lost confidence.
Among UK businesses whose AI implementations have failed to meet expectations, 24% cite inadequate network or connectivity performance as a contributing factor. That figure matters because it shows how easily an ambitious AI project can become an expensive pilot when the foundations are not ready.
The network has moved into the ROI calculation
For years, networks were treated as background infrastructure. That was easier to justify when the workload was email, file sharing or standard SaaS access. It is much harder to defend when AI is embedded into live operations, customer workflows, fraud detection, forecasting or field service.
AI creates a different operating requirement. The business needs to move data securely, route workloads intelligently, maintain uptime and understand where cost is accumulating. Without that visibility, ROI becomes reactive, without resilience AI-enabled processes remain fragile – and without secure connectivity, new use cases create new points of exposure. This is why the infrastructure conversation has to take place much earlier in the AI lifecycle.
Looking ahead, 54% of UK organizations say they need more flexible and scalable networks to thrive in an AI-driven environment, while 57% say they need greater resilience and reliability to maximize uptime.
Those numbers should not be read as a technical wish list, they are signals that AI is forcing enterprises to rebuild the conditions around digital performance. If AI is expected to support decisions, customers and operations, the network underneath becomes part of the growth plan.
Four questions before the next round of AI funding
Before signing off the next AI budget, boards should ask where the data will move, who can see it moving, whether the network can support AI beyond the pilot stage, and what happens when something fails.
Those questions force the board to look below the application layer and examine whether the organization has the conditions to scale what it is funding. They also give the CIO a way to connect ambition to operational reality underneath: infrastructure, skills, governance, security, data movement and cost discipline.
The hardest part of AI ROI is that some value will take time to mature. No board would judge the value of the internet by what it delivered in its first year, and AI should be approached with the same understanding. Short-term measurement matters, but it cannot become an excuse to underinvest in the capability building that makes long-term value possible.
The next wave of AI investment will not be won by enthusiasm alone. Before the next model is selected, the next platform is funded, or the next pilot is announced, UK boards should look again at the network underneath.
AI investment driven by fear of missing out will only carry enterprises so far. Durable value will come from building the right conditions around it from the start.
We've featured the best network monitoring tools.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit