A Hurricane opened my eyes.
Hurricane Seen From Space. (NASA)
When Hurricane Milton ended his visit to the place where I live, there was no running water. We prepared beforehand by buying bottled water. We bought bottled water even though we were (and are) paying taxes to have water, because government water systems could not be expected to provide water right after a hurricane passes through.
When Hurricane Milton came through here, there was little or no electricity. Many people were without it. I learned from experience during Hurricane Charley and I was fortunate enough to buy a generator. We pay to keep it filled and maintained all through the year, not just when a hurricane approaches.
Buildings were converted into hurricane shelters for people. Many people sheltered from the storm in these large spaces under a roof and four walls. Hurricane shelters were usually opened in large spaces because many people would be there and each would get a small space to shelter from the storm.
Experiencing the aftermath of Hurricane Milton opened my eyes to many things. One of them is that the aftermath of Big Tech is like the aftermath of Hurricane Milton with Nondisclosure Agreements.
“We’re past the details I can talk publicly about.” CEO of a startup intended to generate electricity out of fusion.
The biggest thing about Big Tech is secrecy. Artificial Intelligence or “AI” offers the prime example, but Cryptocurrency is another. This article will concentrate for now on Big Tech and AI. AI has literally changed the landscape with its enormous drain on the energy grid and on water resources.
AI is a high-tech covert operation. Secrecy is embedded in the program every bit as much as “1s” and “0s” are embedded in the computer programming. In another forum I previously wrote about the use of Nondisclosure Agreements required by Big Tech companies from the utilities that provide them with the energy to run their AI operations and from the local governments that make their AI operations possible. As I mentioned in an earlier Substack newsletter, utilities and government officials and agencies sign nondisclosure agreements (NDAs) with Big Tech to keep the deals strictly quiet from the people who live there.
AI is housed in computing factories called “Data Centers.” Data centers have existed for decades, housing computer servers that have been engaged in things from email to commercial activities. The computers are set out in what have been called “factory-libraries” where they are contained in stacks, like books in the stacks at a library.
Construction_on_the_Boston_University_Center_for_Computing_&_Data_Sciences_
(51568555756). (Wikimedia Commons)jpg
Heat is a feature of data centers, not a bug, as they say. The workers in them are few and their main task seems to be to avoid “a thermal runaway event,” a “thermodynamic threat.” For that reason, the cooling agent of choice is overwhelmingly air conditioning. Air conditioning is used to cool the computer servers that generate such heat.
DATA LOGGER.jpg (Wikimedia Commons;
Data logger used to monitor temperature and relative humidity)
Although data centers seem to have (practically) always been with us, AI has changed them by causing exponential increases in the consumption of electricity and water. Today, one data center can equal the electricity of some 50,000 homes. Another data center, near Omaha, Nebraska, uses the same amount of electricity as roughly half the homes in Omaha.
Collectively, since AI came along, data centers in this nation are using more energy than some countries. Not this nation. Just the data centers that are in this nation. The point is clear, I hope: We are talking about numbers in the stratosphere, numbers that are beyond huge.
Data centers consume energy because they serve a larger purpose for Big Tech: The purpose is to consume energy to prevent costly downtime in the delivery and use of AI. For this reason, the centers usually depend on backup diesel generators that are ready to turn on the moment that other energy sources unexpectedly become unavailable or, to put it another way, “go offline” – just like the reason that I have a generator at my house because of hurricanes like Milton.
For that matter, coal-burning plants are left online to accommodate the greater demand for energy since AI.
A phenomenon, which probably happens more often than other things that are called phenomenons, is “flood cooling,” which means preventing thermal runaway events by making the center cold regardless of the cost and regardless of the means.
That is where water comes in. Data centers use huge amounts of electricity because electricity powers things like air conditioners that cool the computers as they do their work. Similarly, data centers use water at tremendous levels and for the same purpose: cooling the computers.
Airship_flying_over_the_Utah_Data_Center_04.jpg
(Greenpeace; Wikimedia Commons)
Some examples may illustrate the enormous drain on water supplies that data centers require. The National Security Agency’s Utah Data Center uses 7 Million Gallons of water a day, while the nearby residents of Bluffdale, Utah suffer from water shortages and power outages.
An EMail of 100 words generated via AI once a week for a year by 1 out of 10 Americans equals the same amount of water that every household in Rhode Island will consume in a day and a half.
Parenthetically, who among us doesn’t write emails at work if we have access to a computer and the internet on or as part of our job?
______________________________________________
The biggest objection to the way that data centers are used today is that Big Tech is failing in keeping its climate pledges.
“We all want to be cleaner. But you guys aren’t going to wait 10 years.” President of a utility speaking at a recent gathering of data center executives.
Big Tech is failing as a result of the way it runs these data centers for quick profits from AI vs. the “clean energy” solutions over the long term to which they previously committed themselves. And from which they are now very quickly backing away, in favor of making high profits now. AI demands more energy, right now.
The choice faced by Big Tech in the current operation of data centers to accommodate AI is a choice between cost and the environment, short- vs. long-term. The choice involves consequences beyond breaking promises on clean energy. It may extend to taking over existing and potential sources of electricity.
In the end, the choice is not Big Tech’s alone to make. It is ours.
—
Notes:
Evan Halper, A Utility Promised to Stop Burning Coal. Then Google and Meta Came to Town, WASHINGTON POST (online Oct. 12, 2024).
Evan Halper and Caroline O’Donovan, AI is Exhausting the Power Grid. Tech Firms Are Seeking a Miracle Solution, WASHINGTON POST (online June 21, 2024). The quotes from the startup CEO, “We’re past the details I can publicly talk about,” and “We all want to be cleaner. But you guys aren’t going to wait 10 years,” from the utility president speaking at a meeting of data center executives, came from this reporting.
Steven Gonzalez Monserrate, The Cloud is Material: On the Environmental Impacts of Computation and Data Storage, MIT SCHWARZMAN COLLEGE OF COMPUTING (Winter 2022; online Jan. 27, 2022). This study is a landmark in the literature on the role of Artificial Intelligence in Big Tech’s data centers. The author actually visited data centers as part of his research and his personal observations enrich what might have otherwise been a rather dry study immensely.
THE INK, The Island Where Three Apocalypses Meet (online Sept. 25, 2024).
Pranshu Verma and Shelly Tan, A Bottle of Water Per Email: The Hidden Environmental Costs of Using AI Chatbots, WASHINGTON POST (online Sept. 18, 2024).
Please read the disclaimer. ©2024 Dennis J. Wall. All rights reserved. Interested in many things including Claims and Issues? Check out my Substack newsletter.