Powering the AI Era

Goldman Sachs Investment Banking examines the massive power and capital requirements of AI-era data centers, projecting a 160% surge in global data center power demand by 2030. The report highlights how AI server racks consume 10x more power than cloud equivalents, creating a critical grid bottleneck that natural gas (5–7 year lead times) and intermittent renewables cannot quickly resolve. It argues that an "all-in" approach combining multiple power sources, behind-the-meter solutions, and nuclear energy will be needed, alongside innovative financing to fund the roughly $12 billion cost of a typical 250 MW AI data center.