They are planning on building a 10 GW data center here in Cheyenne, Wyoming, and generating all of the electricity on site..
“Wyoming is poised to become an artificial-intelligence powerhouse after Laramie County commissioners last week unanimously voted to move forward with the construction of a 1.8 gigawatt data center designed to eventually scale up to 10 gigawatts, which would be the largest single AI campus in the U.S. “
10 GW is the equivalent of one hundred million 100W light bulbs.
I asked Google Gemini to calculate the energy balance.
To calculate the true environmental heat flux of a 10 GW compute footprint, we must look at both sides of the meter: Silicon Dissipation and On-Site Generation Inefficiency.
The Silicon Compute Load (100% Thermal Conversion)
According to the First Law of Thermodynamics, energy cannot be destroyed. Inside the computing cluster, electrical energy is fed into transistors to drive bit transitions. Aside from a negligible fraction of a percent that escapes as light through fiber optics, 100% of the computing wattage degrades into Joule heating due to electrical resistance.
The mathematical representation of continuous heat generation (E_compute) over one hour (t = 3,600 seconds) is:
E_compute = P * t
E_compute = (10 * 10^9 J/s) * 3,600 s = 3.6 * 10^13 Joules (36 Terajoules)
The On-Site Generation Load (50% Thermal Inefficiency)
To deliver 10 GW of usable electricity to the servers, an on-site power plant must burn fuel. If the plant operates at a standard 50% thermal efficiency (eta = 0.50), it means that for every 1 Joule of electricity produced, 1 Joule of raw heat is lost directly to the environment through the plant’s own cooling towers and exhaust stacks.
The total fuel energy input required (P_input) is:
P_input = P_compute / eta = 10 GW / 0.50 = 20 GW
Thus, the heat wasted purely by generating the electricity on-site (E_generator) over one hour is:
E_generator = 10 GW * 3,600 s = 3.6 * 10^13 Joules (36 Terajoules)
Total Combined Heat Footprint
Adding the heat from the chips to the waste heat from the on-site generators gives us the total hourly thermal output (E_total):
E_total = E_compute + E_generator
E_total = 3.6 * 10^13 J + 3.6 * 10^13 J = 7.2 * 10^13 Joules (72 Terajoules)
Quantifying the Yield of an Atomic Bomb
To provide context for 72 Terajoules, we examine the energy released during a nuclear explosion. The “Little Boy” weapon detonated over Hiroshima was a uranium-235 fission device. In physical reviews, the energy metrics of nuclear weapons are calibrated in kilotons (kt) of TNT equivalent.
By international scientific convention, one gram of TNT releases exactly 4,184 Joules. Therefore, a standard metric kiloton of TNT translates to:
1 kt TNT = 4.184 * 10^12 Joules
Forensic assessments by Los Alamos National Laboratory place the explosive yield of the Hiroshima detonation at approximately 15 kilotons. To determine the absolute energy yield (E_bomb) in Joules, we compute:
E_bomb = 15 kt * (4.184 * 10^12 J/kt) = 6.276 * 10^13 Joules
The Hourly Comparison: Side-by-Side Analysis
We can now calculate the explicit ratio (R) between the hourly heat emission of our 10 GW megastructure (with on-site generation) and the total yield of the Hiroshima bomb:
R = E_total / E_bomb
R = (7.2 * 10^13 J) / (6.276 * 10^13 J) = 1.147
Metric Comparison:
Energy Release for 10GW Data Center Complex (Every Hour): 7.2 * 10^13 Joules
Energy Release for Hiroshima Atomic Bomb (Total Yield): 6.276 * 10^13 Joules
TNT Equivalent for 10GW Data Center Complex (Every Hour): 17.2 Kilotons
TNT Equivalent for Hiroshima Atomic Bomb (Total Yield): 15.0 Kilotons
When factoring in the 50% thermal efficiency of on-site generators, a 10 GW data center campus generates 114.7% of the total energy of the Hiroshima bomb every single hour. Stated another way, the facility matches the energy footprint of an atomic bomb every 52 minutes.
A study done of existing smaller data centers shows a massive data heat island effect extending many kilometers away.
(PDF) The data heat island effect: quantifying the impact of AI data centers in a warming world
Steel plants generate a lot of heat, but are located near bodies of water which dissipate the heat and minimize any local temperature increases. Without water to transport and evaporate the heat away, the air is going to become very hot.
From Gemini:
“Because water’s volumetric heat capacity is roughly 3,500 times greater than that of air, a tiny temperature shift in a body of water represents a massive reservoir of energy. Mathematically, transferring the exact amount of thermal energy required to raise a specific volume of water by just 2°C into an identical volume of air would trigger an astronomical temperature spike of nearly 7,000°C—well beyond the 6,000°C mark.
To put this into perspective, this concentrated thermal energy would make that volume of air hotter than the radiant surface of the sun, which burns at approximately 5,500°C.”
The Rawhide Generating Station near us in Colorado uses water to evaporate away the heat. I took this picture at -10F on February 14, 2021 and you can see the water evaporating off the cooling pond and out of the stack.
Evaporating water is the only viable way to remove large amounts of heat from the local environment, but these data centers are being sold based on the idea that they won’t use water.
We are having the warmest and driest year on record here in Cheyenne, and there is no snow at any of the locations we depend on for our water supply.
https://www.ncei.noaa.gov/pub/data/ghcn/daily/all/USW00024018.dly
Besides the noise, pollution and heat, this will greatly increase the competition for energy, which will drive up fuel prices.
In order to house the workers, they are building hideous four story apartment complexes all over town.
Report projects worsening housing crisis in Cheyenne | Local News | wyomingnews.com
Our community and environment are being destroyed by these data centers, but big money seems to get whatever they want. “Consent of the governed” no longer seems to be a valid concept in America.
“Governments are instituted among Men, deriving their just powers from the consent of the governed”
