The AI revolution is a power-hungry one. The $3 trillion spending spree on AI datacenters, which now number 11,000 globally (a 500% increase in 20 years), is creating an “exponential demand” for energy. This demand is so great that it will require a $720 billion investment in grid spending, according to Goldman Sachs.
The scale of the new “central nervous system” for AI is massive. Global datacenter capacity is 59GW, but Goldman Sachs expects this to double by 2030. This year alone, 10GW of new capacity is set to begin construction—a power draw equivalent to one-third of the UK’s entire grid.
This $3tn build-out is being led by “hyperscalers” like Google and Microsoft, who are spending $750bn in two years. Their projects, like the new datacenter in Newport, Wales, are “general purpose” facilities, powering AI alongside everyday email and Zoom calls.
However, the “speculative” side of the boom, which is “all-in on AI,” is also contributing to this energy demand. Mega-projects like the $500bn “Stargate” are being planned.
While analysts debate if the $3tn spend is a “bubble,” the $720bn grid cost is a hard physical and financial necessity. The AI boom is not just a digital shift; it’s a massive energy and infrastructure project that is testing the limits of our global power grid.