AI infrastructure at a crossroads: why holistic data center design can’t wait (www.techradar.com)

AI-driven demand has pushed data centers to a crossroads: operators report a rapid, uneven surge in AI workloads (92% saw demand grow, by an average 42%), while geopolitics, tariffs and site scarcity (69% and 40% respectively) are delaying builds and driving up costs. Many say demand outstripped expectations (64%) and will shorten facility lifespans (58%), prompting 74% to rethink power, cooling, site and architectural strategies. National constraints are already reshaping plans—12% of Irish projects paused for energy concerns versus 5% in the UK and 2% in the Nordics—while the UK’s designation of data centers as Critical National Infrastructure and £6.3bn in global investment underline the sector’s strategic importance. The technical and operational implications are stark: AI workloads require far more power, cooling and network throughput than legacy designs, and common oversights—especially cabling—threaten long-term performance (70% flag poor cabling as a critical risk). Skills and partner shortages (≈80% and 28%) further imperil projects. The remedy is holistic, forward-looking design: define purpose, align stakeholders, build modularity and adaptability, use AI for optimization (energy, performance, predictive maintenance), and avoid false economies by investing early in quality materials and systems. Operators who integrate power, cooling, cabling, site selection, sustainability and talent planning from the outset will be best positioned to deliver resilient, efficient AI-ready facilities.