More Than Two-Thirds Of The Power Sought For US Data Centers Will Never Materialize
Some developers are pitching the same project to multiple utilities, with plans to push ahead with the application that lands the best deal and speediest approval.
To crack down on the barrage of requests, many utilities have introduced steep upfront application costs, demanding big-money collateral and near-perfect credit ratings. Small to mid-sized developers often pay project costs up front and then sell completed facilities to well-capitalized AI companies.
"Utilities are using what they'll refer to as a 'first-ready, first-served' model, as opposed to a 'first-come, first-served' model, to weed out folks that really don't have the capability to deliver".
https://www.zerohedge.com/energy/most-two-thirds-power-sought-us-data-centers-will-never-materialize
AI:
Artificial intelligence data centers require massive amounts of electricity, with global data center consumption projected by the International Energy Agency to double from roughly 415–448 TWh to nearly 945 TWh by 2030. Large hyperscale AI facilities frequently demand 100 megawatts or more, matching the continuous power consumption of 100,000 households. Traditional data center racks require 10–15 kilowatts, whereas AI-focused server racks packed with specialized processors require 50–150 kilowatts. Individual high-performance AI graphics processing units (GPUs) consume 700 to 1,200 watts each, compared to 150 to 200 watts for standard server CPUs. High-density heat output requires intensive cooling and ventilation infrastructure, consuming anywhere from 7% to over 30% of total site energy. Initial deep-learning model training requires massive continuous clusters running for months, while everyday AI inference handles billions of routine user prompts globally.
You might remember before AI rolled out about how media wondered how we'd have enough electricity for the projected consumption of an all EV fleet of consumer driven cars.