We’ve got three primary obstacles. The first is generation capacity. Assuming for the moment that the data center boom ends today. The United States needs to expand its grid by approximately 50%. And that’s obviously a lot. It took a half century for us to do the last 50% expansion. So to do that in a short period of time is going to be very, very difficult.
With the data center expansion. You’re talking about a doubling or tripling or a quadrupling. And so anyone who tells you that data centers are the wave of the future clearly hasn’t done the math.
An LPT is basically something that is on the grid rather than in the power plant, and it steps up and steps down the electricity for transmission and distribution.
They weigh 100 to 400 tons. And it’s a perfect example of one of those mid tech but high skill industries and products that the United States is basically outsourced. Very, very few of them are made in the United States at all. We import about 80% of what we use and get this. The average transformer, the average LPT in the United States right now that’s operational is about 40 years old.
And right now, there’s a backlog of 3 to 5 years based on which model you’re looking for.
And these things typically last about 40 years. So not only do we need at least 50% more, probably more than that. We also have to overhaul and replace the entire grid supply of LPTs. The volume required here is immense. And the problem is that not a lot of places make these things at scale.
It appears we have trouble here.
Perhaps waterfall can weight in regard to Zeihan’s claims.
I can’t vouch for everything Zeihan says in the video, but he is usually pretty good with the facts and figures. The 3 to 5 years lead time appears to be for the large step-up transformers at the power plants. (They step-up the voltage at the power plant, transmit it some distance on high voltage wires, and then step-down the voltage at the substations and other local transformers, by the time it gets to your house.)
There seems to be a general consensus that more data centers are needed, and those data centers consume a lot of power. Also, many people think EVs are going to be a big thing in the future, and those EVs will need to get their charge from somewhere. The basic math says we need more electrical “stuff” for a larger grid handling more and more gigawatts each year. Transformers are an important part of that “stuff”.
I actually posted about this last year.
A quick Google search seems to indicate the lead times aren’t getting much better this year.
The problem is that the power source is far from the data centers. The grid worked fine for big power sources and smallish power users. Generate power near the data centers to solve the problem. Small nuclear reactors? Solar?
In the article I gift linked yesterday, you can see many of the [huge] new data centers are putting their own power sources adjacent to the centers. A few with solar, many with gas turbines. Some have been squeezed right into neighborhoods, so there’s no place place to get the juice except from the grid - but I can’t imagine what the noise would be like living right next to those immense fans used to push out the heat.
It seems rather counterproductive to champion the use of EVs and call for the reduction of fossil fuels, while at the same time pushing out billions of BTUs of heat to power data centers so people can say “Draw a picture of me as a dog.”
The article is correct about tranformers being a limiting items for data centers. But it ignores the big problem of building of transmission lines needed to export the electricity from the generation (swithchyard and transformer) to the data center (switchyard and transformer). That is major limiting factor for data centers.
I have some very big concerns with the whole Big Tech industry’s rush to build data centers as listed below:
Where is the money coming from? Is Big Tech borrowing hundreds of billions of dollars to finace the engineering, licensing, construction, and procurement of the following?
data center land, buildings and internal electrical gear with difficult licensing with water usage and noise.
power generation facilities with difficult licensing due to safety, pollution, water usage, and difficult produrement of long lead mechanical and electrical components.
transmission facilities with difficult licensing routes and procurement of long lead electrical components.
Dr. Matthew Gardner, vice president of Power Generation System Operations at Dominion Energy, the clearer emerging challenge relates to translating rapid load growth into generation, transmission, substations, and other infrastructure that can take years to plan and build.
“As an energy infrastructure provider, I cannot make plans to serve you what you need today,” Gardner said. “This infrastructure is decades in the making.” He pointed to one 500-kV transmission project that took 11 years to move from concept to service.
The problem with your solution is that people do not want these data centers, power plants, transmission lines, and substattions in their neighborhoods or in their cities. But Big Tech wants their data centers to be close to the people and their cities.
