CRH執行長談AI與資料中心策略

• Data center utilization rates have improved significantly from late 1990s telecom fiber glut (3% utilization) to current high-capacity operations, with Microsoft reporting data center shells nearly at full capacity[1] • AI demand measured in tokens: Google consumed 980 trillion tokens in June 2025 against global capacity of 2,000 trillion tokens per month, indicating 49% utilization of current infrastructure[1] • Multi-year supply agreements becoming standard practice, spanning site preparation through aggregates, cement, and concrete foundation work[2] • CRH's research initiatives include hydrogen use, CO2 mineralization projects, novel cements, AI technology integration, and Carbon Capture Usage and Storage (CCUS) systems[6] • Data center power demand projected to increase 160% according to Goldman Sachs Research (2024), driving demand for specialized cooling and infrastructure materials[2]

CRH's positioning in the AI data center infrastructure buildout represents a structural shift comparable to the Interstate Highway System era. The company's 2030 targets (7-9% annual revenue growth, 22-24% adjusted EBITDA margins, >100% free cash flow conversion) suggest management confidence in sustained multi-year demand[4]. However, success depends on: (1) continued hyperscaler capital expenditure at current $690B+ annual rates; (2) ability to secure long-term supply contracts before competitors; (3) execution on specialized product development for AI-specific infrastructure requirements. Smaller regional suppliers face structural disadvantage, while CRH's integrated model creates defensible competitive position. Macroeconomic risks include interest rate volatility and potential trade policy shifts affecting reindustrialization trends[5].