New research from data centre infrastructure specialist Onnec has found that surging demand for AI is heaping pressure on to data centre operators to accelerate builds – increasing the risk of delays, higher costs and remediation after go-live.
The survey of 300 senior decision-makers at data centre operators in the UK, Ireland and Nordics found that 92% of operators say they are being forced to compress build timelines to keep pace with AI demand. But 75% say speed-to-market pressure is forcing design decisions before infrastructure requirements are fully understood.
As a result:
- Safety and cost pressures are rising: 74% say the pressure to accelerate data centre builds is increasing health and safety risks on site, while 67% say their AI-ready build or retrofit costs have increased in the past 12 months. Among those reporting a rise, costs have increased by an average of 42%.
- Quality control is being squeezed: 45% say compressed timelines have reduced time for testing, commissioning or quality assurance, while 43% say they have increased risk of quality issues or rework on completed builds. A further 43% say infrastructure has required upgrades or remediation after go-live.
- Supply chain disruption is delaying delivery: 61% have had a project delayed by supply chain issues, while 79% agree that geopolitical instability will affect the cost or availability of data centre components.
“Operators are under enormous pressure to deliver AI capacity quickly, but speed and readiness are not the same thing,” comments Matt Salter, Global Head of Data Centres at Onnec. “A data centre can go live on time and still need optimisation for the AI workload it was built to support. Reworks, retrofits and remediation are increasingly common. And network infrastructure is often where problems show up first. No amount of compute or GPU power can compensate for infrastructure or cabling that wasn’t designed to keep pace with today’s AI demands.”
Supply chain bottlenecks are holding back delivery
Almost half (45%) of operators are leaning towards new builds to deliver AI infrastructure, but one of the biggest pressure points impacting new builds is the supply chain. While 87% of operators are confident, they have the right supply chain ecosystem to support fast, reliable delivery, delays are still occurring across the whole data centre environment.
Among those who have delayed a project due to supply chain issues, operators reported issues sourcing GPUs and compute (53%), cooling systems (45%), specialist staff (45%), power distribution equipment (43%) and cabling (39%).
Meanwhile, 29% of operators are leaning towards retrofitting to expand AI infrastructure and maintain pace without sacrificing design quality. But retrofitting is rarely a simple fix. Key challenges include cooling limitations for high-density racks (35%), retrofitting live environments without disruption (33%) and insufficient power capacity (30%).
“The operators who come out ahead in the AI race won’t be the ones who moved fastest,” continues Salter. “They’ll be the ones who took a holistic approach to design from the outset, treating cabling, power, cooling and compute as one connected system rather than separate workstreams managed under deadline pressure. Without that, operators risk locking in performance constraints that are harder and more expensive to address after go-live.”
