AI Power Crunch: Industry Faces $220B Debt Burden
The AI industry is facing a significant challenge in meeting its expected power demand, with Gartner predicting that world datacenter power demand will reach 132 GW by 2026 and Morgan Stanley expecting US datacenter demand to go as high as 74 GW by 2028.
Teravolt, a London-based AI infrastructure company, foresees AI demand requiring 410 GW by 2036, a 240 GW shortfall from the expected availability of 170 GW from the grid at that time.
The company's plan to bridge this gap involves repurposing existing energy assets like old thermal power plants and industrial sites, which can be converted into datacenters in one to three years, whereas building new grid infrastructure takes five to 15 years.
Teravolt's business model is focused on retrofitting these existing assets, which has been done successfully with Bitcoin mining operations. The company argues that repurposing legacy power generation infrastructure like coal and gas plants will be maintained longer than planned.