The $3 trillion AI datacenter spending spree is not just about adding more servers; it’s about building entirely new, colossal structures dedicated to artificial intelligence. These mega-projects, with names like “Stargate” and “colossus,” illustrate the sheer scale of the industry’s ambition.
One of the most ambitious is the “Stargate” venture, a $500bn joint project between OpenAI, Oracle, and SoftBank. It aims to build a massive network of AI-dedicated datacenters across the US, with a UK version already planned for North Tyneside.
Meanwhile, Elon Musk’s xAI is constructing its own “colossus” project in Memphis, Tennessee, designed to be one of the world’s most powerful AI supercomputers. Not to be outdone, Microsoft is building what it calls the “world’s most powerful AI datacentre” in Fairview, Wisconsin, in addition to backing AI-dedicated sites in the UK.
This new wave of construction is massive. This year alone, 7GW of new datacenter capacity will be completed, with work starting on another 10GW. This is adding to a global capacity of 59GW, which Goldman Sachs predicts will double by 2030. This build-out is so vast it will require an estimated $720bn in new grid spending just to power it.
While these massive, AI-specific projects capture the imagination, they are also the focus of “bubble” concerns. Skeptics, like the chair of Alibaba, point to these huge projects, sometimes funded without customer commitments, as signs of excess. The industry is betting trillions that these “Startgates” and “colossuses” are the necessary factories of the future.