Power requirements are so large, our antiquated grid would require a major upgrade. Plus grid was designed for coal fired power plants. Transition to green energy requires connections to different locations. Delayed grid connections are common. Major grid upgrade required. Routing remains a problem. Right of way can be contested.
Data centers have increasingly large energy requirements. They are probably better off to build their own power plants rather than pay a utility as a middleman. Better reliability is one advantage of the grid–diversified power sources–but at higher cost and likely delays to get connections (and upgrade the grid and build new utility power plants).
Nuclear power (Small Modular Reactors) is an obvious solution but they are probably 10 years away. Wind and solar could work (with battery back up) especially where grid connections are delayed. Natural gas fired power plants are the likely source for now.
Who will supply them. They say GE Veranova is sold out with a two year wait. Google says “Major manufacturers of gas turbines for data centers include GE Vernova, Siemens Energy, Mitsubishi Power, and Baker Hughes.”
Grace Hopper of the Navy carried around an 11.8 in piece of wire as a reminder this is a nanosecond in data transmission. Data centers should ideally be located near major users to minimize delays.
Ask Google how many data centers are under construction and you will get an ear full. The US has plans for abt 3000 data centers with abt 475 under construction–with newer ones much larger.
The ideal location is where energy (natural gas), cooling water, and population centers come together. Most are in silicon valley, northern Virginia, and Dallas-Ft Worth. Sources mention Texas, Louisiana, Pennsylvania, Ohio as preferred sites but also Tennessee (for TVA), and Wisconsin (why?). Oklahoma might be considered for natural gas but water is not abundant. Some say the southern tier states along I-20 are attractive (Florida, Georgia, Alabama, Mississippi). Any state with surplus power remains to be considered–at least until the surplus is promised.
Pipelines can carry the gas long distances but permits can delay construction. Adding capacity to existing pipeline right of ways is likely easiest. Marcellus shale in western PA is an especially attractive location.
And don’t forget Canada. Abundant hydropower and water in Ontario and natural gas in Alberta should make them a player.
Meanwhile the anti-data center forces are building. Not in my backyard. But major contributions to tax base can reduce property tax and 30 or so good paying jobs can be attractive in smaller communities. Voters need to decide if data centers are worth the risk.
For now it is likely data centers will be built where ever they are welcome. Financiers wonder if they can make enough money to be profitable. We shall see.