UK data centre sector doubts AI-ready infrastructure

Fluke has published research showing that data centre professionals have low confidence in the accuracy of infrastructure testing data. The findings also highlight widespread concern over whether the UK can support its AI ambitions.

The survey of more than 150 data centre professionals found that only 22% fully trust their test and measurement data to reflect real-world operating conditions. Confidence fell to 19% when respondents were asked about peak load or failure scenarios.

That lack of confidence appears to be affecting day-to-day operations. Half of respondents said they experience unplanned outages or major performance disruptions at least once a year, while 10% reported monthly incidents and 8% said disruptions occur weekly.

Legacy equipment was a recurring concern. Nearly two-thirds of respondents, or 65%, said outdated testing tools increase the risk of downtime and compliance failures within their organisations.

Monitoring Gaps

The research also pointed to weak visibility across core systems. While respondents broadly agreed that regular maintenance is important for reducing downtime, only 28% said they have real-time or predictive monitoring across critical infrastructure such as power, cooling and networks.

One in five said maintenance is carried out no more than quarterly. Adoption of automation, AI diagnostics and predictive monitoring also remains limited, with only 10% saying those systems have been fully implemented. A further 22% said such tools were in pilot programmes, while 19% described deployments as being at an early stage.

Skills shortages emerged as the main reason for poor confidence in infrastructure data. Some 43% of respondents cited skills and training gaps as the biggest barrier, ahead of time pressures during commissioning at 16%, inconsistent testing processes at 11% and budget constraints at 10%.

The findings suggest operators are being squeezed between rising demand and operational discipline. Forty-two per cent of respondents said time pressure creates occasional compliance risks, while 17% said it makes it significantly harder to meet changing connector and certification requirements.

AI Pressure

The results come as AI-related demand adds to existing strain on data centre infrastructure. Operators are being asked to expand capacity while maintaining uptime, testing discipline and regulatory compliance.

Against that backdrop, only half of respondents said the UK data centre sector is operationally ready to scale for AI, cloud and hyperscale demand over the next five years. Just 7% said the UK currently has the infrastructure resilience and operational standards needed to support its stated ambition of becoming an AI leader, while 28% pointed to significant infrastructure gaps.

The headline figure in the wider survey was even starker: 93% of professionals believe the UK lacks the necessary infrastructure to support those ambitions.

The responses reflect a sector facing both physical and organisational constraints. Demand for denser computing environments, more complex fibre networks and tighter performance requirements is increasing, but many operators still appear to rely on older testing and maintenance methods.

Mike Slevin, Director of EMEA Market at Fluke, said the issue is not a lack of understanding about the need for better processes.

"What's striking here is that organisations already know what needs to be done. There's broad recognition that regular maintenance and better monitoring are critical to reducing downtime, yet in practice, adoption is lagging," he said. "That gap between awareness and action is where risk builds. When testing isn't consistent and monitoring isn't real-time, small issues can quickly escalate into outages."

The survey was conducted among global data centre professionals at Data Centre World London. It asked 11 questions on infrastructure confidence, data accuracy under real-world conditions, operational risk, and testing, monitoring and maintenance practices.

Slevin said the technical demands created by AI workloads are narrowing the margin for error in data centre operations.

"AI is redefining the demands placed on data centre infrastructure. With higher-density architecture and increasingly complex fibre environments, multi-fibre testing has become paramount as the margin for error narrows," he said. "If organisations can't confidently validate performance under real-world conditions, they risk building AI on unstable foundations. The challenge now is ensuring that capacity is resilient and ready for sustained demand."