#electricity use

[lids-flutter-open]

Data Centers and Water Consumption | Article | EESI

Once again begging people to note that AI consumes water more intensely than prior data center uses; while it is not the #1 drain on water (that is stuff like beef production and industrial production) it is rapidly increasing water demand as more companies build data centers. This is putting a major strain on water tables and river ecosystems which we cannot afford. The colorado river is one adjacent to the biggest boom in AI and AI/expanded data centers for cloud computing, but my home state of Washington is also seeing massive demand along the Columbia, which has been the subject of a salmon population crisis for decades and which had recent climate wins that were then reversed:

Salmon, tribal sovereignty, and energy collide as US abandons Resilient Columbia Basin Agreement

The move threatens treaty rights and salmon recovery as energy demands from AI and crypto surge.

This free article is recommended reading for anyone who says that AI is not driving policies that are harmful to our ecosystems. Even if AI is a small fraction of data center electricity use and most of it is cloud computing, demand is increasing in a way that has led lobbyists to urge Trump etc to reverse climate commitments toward taking out dams and restoring fish populations.

As drought deepens, big tech has put nearly half of its data centers in water-scarce regions

The largest data centers can guzzle millions of gallons of fresh water a day. Roughly 40% are located in the most water-stressed areas of th

https://andthewest.stanford.edu/2025/thirsty-for-power-and-water-ai-crunching-data-centers-sprout-across-the-west/

Thirsty for power and water, AI-crunching data centers sprout across the West

With promises of jobs and hopes for tax breaks, server farms are reshaping local grids, plumbing, and politics. Are they a boon for communit

https://gerrymcgovern.com/why-do-data-centers-love-deserts/ data centers also use water that could otherwise be used by people, and use it by evaporating it, making it unavailable in the same watershed in a way different than individual consumer use. This is driving up water bills in desert communities. Data centers often get water extremely cheap, competing with tribes, agriculture and consumers for water, but pay a tiny fraction of what individuals pay for water. They also often do not even measure their own water use:

https://gerrymcgovern.com/big-tech-lies-about-its-water-use/

Big Tech lies about its water use

Why do data centers love deserts?

[rjzimmerman]

The U.S. broke electricity demand records twice in July (Heatmap AM)

The United States’ demand for electricity broke records twice last month. Air conditioners cranking on hot days, combined with surging demand from data centers, pushed the peak in the Lower 48 states to a high of 758,053 megawatts on July 28, between 6 p.m. and 7 p.m. EST, data from the U.S. Energy Information Administration’s Hourly Electric Grid Monitor shows. The following day, peak demand set another record, hitting 759,180 megawatts. That’s nearly 2% above the previous record set on July 15, 2024. 

The EIA predicted demand to grow by more than 2% per year between 2025 and 2026. Forecasts are even higher in areas with large data centers and factories underway, such as Texas and northern Virginia. The milestone comes as the Trump administration cracks down on solar and wind energy, two of the fastest-growing and quicker-to-build sources of new generation. On Tuesday, The New York Times reported that the Environmental Protection Agency is moving to eliminate $7 billion in spending on grants for solar energy, though when Heatmap’s Emily Pontecorvo asked the agency, it said only that, “With the passage of the One Big Beautiful Bill, EPA is working to ensure Congressional intent is fully implemented in accordance with the law.”

[jgkoomey]

Our new article published today in Joule: “Does not compute: Avoiding pitfalls in assessing the Internet's energy and carbon impacts”

I and Professor Eric Masanet of UC Santa Barbara have a new commentary article out today in the refereed journal Joule. It explores four common pitfalls that cause researchers and commentators to exaggerate information technology electricity use and emissions, and suggests four ways industry and researchers can avoid spreading such misconceptions in the future.

It’s a short article, so I won’t spoil it by giving too much away, but the figure above summarizes one key lesson from our review: Growth in data traffic in either the short term or the long term does not necessarily imply growth in energy use. It depends on how fast efficiency improves!

I summarize the conclusions in my June 1, 2021 keynote for the iTherm conference:

Here’s the reference:

Koomey, Jonathan, and Eric Masanet. 2021. "Does not compute: Avoiding pitfalls in assessing the Internet's energy and carbon impacts." Joule. June 24. [https://doi.org/10.1016/j.joule.2021.05.007]

[jgkoomey]

I gave a virtual keynote today for the iTherm 2021 technical conference, focusing on misconceptions about electricity use and emissions associated with computing

My keynote talk today for the iTherm 2021 technical conference is an expansion of points made in a commentary article by me and Professor Eric Masanet, UCSB, which is “in press” at Joule right now (more when that’s published). I presented nine different high-profile misconceptions about electricity use and emissions associated with computing, explored four pitfalls that lead to such misconceptions, and suggested four ways we can do better in the future.

Here is the conclusions slide:

You can download a PDF of the slides (which include three pages of references) HERE.

[goingtiny]

I’ve spent the last month trying to reduce the carbon footprint of the websites I have (some) control over. When talking about this with other people they often look at me blankly before asking “aren’t you taking this a little too far?”.

All of my WordPress plugins combined run on well over 2 million different websites, each website receiving who knows how many visitors.

At an average energy expenditure of 0,5 kWh per GB 1 of data transfer this means that every kB equals 0,5 kWh / 1.000.000 kB * 2.000.000 websites = 1 kWh if each of these websites received exactly 1 visitor.

Let’s assume the average website receives about 10.000 unique visitors per month and serves files from cache for returning visitors. The total amount of energy saved by shaving off a single kilobyte is then 1 kWh * 10.000 visitors = 10.000 kWh.

10.000 kWh of energy produced by the current European electricity grid equals about 10.000 * 0,295 = 2950 kg of CO2.

Shaving off a single kilobyte in a file that is being loaded on 2 million websites reduces CO2 emissions by an estimated 2950 kg per month.

To put this into perspective, that is the same amount of CO2 saved each month as:

Driving my Toyota Yaris for 18.670 kilometers. (158 g CO2 per km)

5 flights from Amsterdam to New York. (679 kg CO2 per flight)

Eating 118 kg of beef (25 kg CO2 per kg of beef)

I already work from home, am a vegetarian and didn’t take any flights in the last 3 years so it seems I am stuck trying to make the web more efficient.