On Oct. 16, tech giant Amazon announced it had signed three new agreements with electricity providers to support the development of a number of nuclear energy projects, including the development of several Small Modular Reactors (SMRs) to help power the company’s growing electricity needs, particularly the electricity needed to fuel its artificial intelligence (AI) projects.
Amazon joins other tech giants, including Google and Microsoft, who are each betting big on nuclear energy to satisfy the growing electricity needs of their AI ambitions.
As of the end of 2023, data centers around the globe accounted for about 1 to 1.5 percent of global energy use and the International Energy Agency (IEA) estimates that this usage could double by as early as 2026, according to Popular Science.
Nuclear reactors such as SMRs have “a smaller physical footprint, allowing them to be built closer to the grid,” Amazon says, and compared with traditional reactors, they can be built and deployed much faster.
In a statement, Matt Garman, CEO of Amazon Web Services, said: “Nuclear is a safe source of carbon-free energy that can help power our operations and meet the growing demands of our customers, while helping us progress toward our Climate Pledge commitment to be net-zero carbon across our operations by 2040.”
Success
You are now signed up for our newsletter
Success
Check your email to complete sign up
Garman added that Amazon believes that nuclear energy is “an important area of investment for Amazon,” because nuclear energy is “both carbon-free and able to scale.”
READ MORE:
- America Bets Big on Nuclear as Coal Plants Wind Down
- US Announces Global Nuclear Fusion Strategy at COP28
- British Official Announces ‘Nuclear Power Renaissance’
Amazon energy partners
Amazon has entered into an agreement with Washington-based Energy Northwest to produce at least four advanced SMRs that will be developed, owned and operated by the public power agency.
Amazon will also be working with X-energy who will be designing some of the SMRs used for Amazon’s projects with Energy Northwest.
Among other stakeholders to the agreements is Virginia-based Dominion Energy, which will be responsible for developing an SMR near its North Anna Nuclear Generating Station.
Artificial intelligence networks are becoming some of the largest energy consumers in the tech sector. With the exponential rise in demand for complex AI models, like ChatGPT, and other AIs that deal in language processing, image generation, and other data-intensive applications, the energy needs of AI continues to surge.
As such, Microsoft announced in September its plans to purchase power from the Three Mile Island plant in Pennsylvania, a plant which is scheduled to reopen specifically to provide energy for Microsoft. Three Mile Island happens to be the site of the worst nuclear power disaster in the history of the United States.
Earlier in October, Google also announced a deal with nuclear power startup, Kairos Power, to buy 500 megawatts of electricity generated by seven yet to be built SMRs.
The added capacity is needed. It’s estimated that a single ChatGPT inquiry can use as much electricity as a light bulb requires for 20 minutes and, in November 2023, the platform received more than 10 million queries per day. The number of daily inquiries has only grown since then.
READ MORE:
- Korean Mother Files Lawsuit Over Forced Adoption of her Daughter to US
- Rare Birth of Endangered Lemur Celebrated in Cotswold Wildlife Park
- New Sino-Indian Border Agreement Signals Thawing Relations, But Questions Remain
Environmental costs
Artificial intelligence is having a significant impact on the climate ambitions of big tech companies due to the amount of electricity and water it uses.
According to a report by the Washington Post, a single 100-word email written by ChatGPT uses the same amount of electricity that would be needed to light 14 LED light bulbs for one hour.
Send the same email once a week for a year and the platform uses enough energy to power 9.3 Washington D.C. households for an hour.
If one out of every 10 Americans uses ChatGPT to create and send one 100-word email a week for a year, the energy consumed would be able to power all D.C. households for 20 days.
In addition to electricity, the amount of water AI platforms use is gargantuan.
For the same 100-word email, more water than would fit in a 500 milliliter bottle is needed to cool the servers.
The energy AI servers use to calculate outputs creates an immense amount of heat, and many servers are cooled primarily by water fed systems.
Sending the same email once a week for a year would require 27 liters of water, or about 7.13 gallons. If one in every 10 Americans did the same thing for a year, it would require 435,235,476 liters, or around 115 million gallons of water.
However, depending on where the data center is located, electricity could be used to cool the servers via industrial sized air conditioners.
That said, even in ideal conditions, data centers are often the largest user of electricity and water in the jurisdictions where they are located, raising concerns that they may drive up costs for residential users in the area.
