Tech
An Amazon Web Services data center is shown situated near single-family homes on July 17, 2024 in Stone Ridge, Virginia
An Amazon Web Services data center in Stone Ridge, Virginia (Nathan Howard/Getty Images)
power hungry

AI data centers are devouring our energy and propping up fossil fuels

A new report finds the explosive growth of AI data centers is consuming a disproportionate amount of our energy generation.

Jon Keegan

The rush to build ever-larger, power-hungry data centers is a drag on America’s progress toward clean energy. And there’s no sign of things slowing down.

Earlier this month, Amazon announced it will be investing $11 billion to build out AI data centers in Georgia, and Microsoft announced it will be spending $80 billion on data centers around the world in FY 2025.

A new report from think tank Frontier Group, public research group US PIRG, and Environment America details this massive power consumption by the data centers leading the AI revolution and cryptocurrency.

The report highlights some startling data that illustrates the massive gap between power supply and demand.

Insatiable demand

In the states where tech giants have been pouring billions of dollars to rapidly build computing infrastructure for AI, the data centers are devouring huge percentages of each state’s power generation.

Virginia, which is home to a huge number of data centers, expended more than a quarter of the state’s total electricity generation to such data centers in 2023. For comparison, in 14 other states, less than 1% of the state electricity generated went to data centers, according to the report.

Another issue highlighted by the report is the huge variance in estimates for future power consumption forecasts. Analysts’ forecasts for growth of electricity demand from 2023 to 2030 varied between 29% to 166%, leaving states with no easy way to ensure enough supply.

Prolonging the use of fossil fuels

The study identified at least 17 fossil-fuel-fired power plants in five states that have moved to delay their phaseouts due to surging demand from data centers. The report identified new fossil fuel plants being planned to help meet demand with at least 10,808 MW of power, all which slow the country’s progress toward transitioning to renewable energy generation.

The report examined the impact on communities adjacent to these supersized data centers and found many significant impacts. The massive use of community water supply, noise pollution, and soaring energy price increases that get passed on to consumers were just some of the impacts.

Negligible societal benefits

The authors included several recommendations to address the concerns raised in the study. Among the recommendations is greater transparency from the tech companies that build these data centers, to better understand energy and water use by the facilities. Other suggestions include requiring new data centers to include on-site renewable energy sources, eliminating public subsidies for data centers, and a deprioritization of computing resources that produce “negligible societal benefits.”

President Trump’s executive order this week titled “Unleashing American Energy” doesn’t contain much to indicate that any of these recommendations will be implemented at a federal level, at least. The administration has made clear that it seeks to streamline the permitting and approvals to generate significantly more cheap energy to power the large AI infrastructure projects its supporting, such as the new $500 billion “Project Stargate” joint venture between Oracle, OpenAI, SoftBank, and its partner Nvidia.

More Tech

See all Tech
tech

Anthropic projections for 2028: Up to $70 billion in revenue, could be profitable by 2027

Anthropic’s Claude API business is doing so well with enterprise customers, the company is upping its revenue forecasts significantly. According to a report from The Information, the company’s robust corporate sales have caused it to revise its most optimistic forecast up to $70 billion in sales by 2028.

Anthropic estimates its API business will be double that of OpenAI’s API sales. OpenAI is currently burning through much more money per month than Anthropic, and reportedly expects to spend as much as $115 billion through 2029, while Anthropic is forecasting that it could be cash positive by 2027, per the report.

Anthropic estimates its API business will be double that of OpenAI’s API sales. OpenAI is currently burning through much more money per month than Anthropic, and reportedly expects to spend as much as $115 billion through 2029, while Anthropic is forecasting that it could be cash positive by 2027, per the report.

tech

Amazon, which is developing AI shopping agents, doesn’t want Perplexity’s AI shopping agents on its site

Amazon has sent a cease and desist letter to Perplexity AI, demanding that it stop letting its AI browser agent, Comet, make online purchases for users, Bloomberg reports.

Amazon, which is developing its own AI shopping agents and is having “conversations” with builders of third-party agents, accused the AI startup of “committing computer fraud by failing to disclose when its AI agent is shopping on a user’s behalf, in violation of Amazon’s terms of service.”

Perplexity, in response, said Amazon is attempting to “eliminate user rights” in order to sell more ads.

Amazon, which is developing its own AI shopping agents and is having “conversations” with builders of third-party agents, accused the AI startup of “committing computer fraud by failing to disclose when its AI agent is shopping on a user’s behalf, in violation of Amazon’s terms of service.”

Perplexity, in response, said Amazon is attempting to “eliminate user rights” in order to sell more ads.

tech

Apple to challenge Google Chromebooks with low-cost Mac laptop, Bloomberg reports

Apple is designing a new sub-$1,000 Mac laptop aimed at the education market, Bloomberg reports.

Google’s low-cost Chromebooks currently dominate the K-12 education market, and Apple’s reentry into the education market that it once owned could disrupt the sectors status quo.

According to the report, Apple plans on using the custom mobile chips it currently uses in iPhones to power the more affordable devices.

Apple’s recent earnings demonstrated that iPhone sales have been steady, and the tech giant is looking to find new areas of growth, like services. A low-cost Mac could be popular with consumers, in addition to education buyers.

According to the report, Apple plans on using the custom mobile chips it currently uses in iPhones to power the more affordable devices.

Apple’s recent earnings demonstrated that iPhone sales have been steady, and the tech giant is looking to find new areas of growth, like services. A low-cost Mac could be popular with consumers, in addition to education buyers.

tech

Getty Images suffers partial defeat in UK lawsuit against Stability AI

Stability AI, the creator of image generation tool Stable Diffusion, largely defended itself from a copyright violation lawsuit filed by Getty Images, which alleged the company illegally trained its AI models on Getty’s image library.

Lacking strong enough evidence, Getty dropped the part of the case alleging illegal training mid-trial, according to Reuters reporting.

Responding to the decision, Getty said in a press release:

“Today’s ruling confirms that Stable Diffusion’s inclusion of Getty Images’ trademarks in AI‑generated outputs infringed those trademarks. ... The ruling delivered another key finding; that, wherever the training and development did take place, Getty Images’ copyright‑protected works were used to train Stable Diffusion.”

Stability AI still faces a lawsuit from Getty in US courts, which remains ongoing.

A number of high-profile copyright cases are still working their way through the courts, as copyright holders seek to win strong protections for their works that were used to train AI models from a number of Big Tech companies.

Responding to the decision, Getty said in a press release:

“Today’s ruling confirms that Stable Diffusion’s inclusion of Getty Images’ trademarks in AI‑generated outputs infringed those trademarks. ... The ruling delivered another key finding; that, wherever the training and development did take place, Getty Images’ copyright‑protected works were used to train Stable Diffusion.”

Stability AI still faces a lawsuit from Getty in US courts, which remains ongoing.

A number of high-profile copyright cases are still working their way through the courts, as copyright holders seek to win strong protections for their works that were used to train AI models from a number of Big Tech companies.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, or Robinhood Money, LLC.