SIAgentNews.com
← All news
DevelopingChipsResearch

U.S. data center emissions exceed those of entire nation of Denmark

A new Harvard study reveals the massive carbon footprint of the AI boom, prompting a new UN pledge for cleaner data infrastructure.

· First reported by The Economic Times on MSN

Blueprint drawing: picture for the story 'U.S. data center emissions exceed those of entire nation of Denmark'
AI-made blueprint drawing for this story. No real people or products are shown.
28.3 million tonsannual CO2 emissions from 403 large U.S. data centers

The short version

  • The 400 largest U.S. data centers emit more CO2 annually than the entire nation of Denmark.
  • More than half of their 82 terawatt hours of electricity use comes from fossil fuels.
  • Electricity use for these centers is enough to power nearly 7 million homes for a year.
  • The Antalya Artificial Intelligence Pledge calls for transparent reporting on water and power use.
  • Data centers could potentially provide enough heat to warm 10% of European homes by 2030.

Why it matters

As AI becomes central to the global economy, its physical infrastructure is creating a significant environmental burden. This story highlights the tension between technological progress and climate goals, showing that the 'cloud' has a very heavy physical footprint.

Measuring the AI footprint

Harvard University researchers recently quantified the environmental impact of the United States' largest computing hubs. By analyzing 403 of the biggest data centers between May 2024 and April 2025, the team estimated that these facilities produce around 28.3 million tons of carbon dioxide every year. This figure is roughly equivalent to the total emissions of a small European country or a state like Michigan.

The study, published in Nature Sustainability, found that these centers consumed 82 terawatt hours of electricity. To put that scale into perspective, that amount of power is sufficient to supply nearly 7 million homes for an entire year. Because data centers are essential for the current artificial intelligence boom, these numbers highlight how rapidly the digital economy is scaling.

The findings were notably higher than previous estimates, with senior author Francesca Dominici stating the results are two to three times what earlier studies showed. Researchers had to use modeling to reach these numbers because many large tech companies do not publicly disclose their specific power consumption or pollution levels. This lack of transparency makes it difficult for policymakers to accurately manage the environmental effects of the growing tech sector.

Fossil fuels and rising demand

A significant portion of the electricity used by these massive data centers currently comes from carbon-heavy sources. The Harvard study revealed that more than half of the 82 terawatt hours of electricity used by the 403 centers came from fossil fuels. These fuels are the primary drivers of global warming, creating a direct conflict with many of the tech companies' stated climate goals.

The demand for data center space is accelerating even as the world tries to reduce emissions. There are currently plans for 2,000 more data centers, which researchers warn could make the existing pollution figures even more worrisome. This explosion of new infrastructure has already caused some major tech companies to struggle with meeting their own environmental targets as their electricity consumption rose 240% between 2022 and 2025.

This rapid growth has led to significant political and local friction. Many communities are pushing back against the construction of new sites due to concerns over energy use and local impact. As the infrastructure expands, the pressure on the power grid increases, forcing a conversation about where this massive amount of electricity will actually come from.

The Antalya AI Pledge

In response to these growing concerns, negotiators at the United Nations COP31 climate talks in Antalya, Turkey, launched a new initiative. The Antalya Artificial Intelligence Pledge is a voluntary agreement for countries and companies to power data centers with clean energy. It also encourages the use of efficient energy and water systems while demanding more openness regarding environmental impacts.

The pledge emphasizes that AI should be designed and managed responsibly by following established environmental standards. It also encourages the reuse or recycling of existing hardware and infrastructure to minimize waste. UN climate chief Simon Stiell noted that we cannot ask people to accept more planet heating pollution in exchange for promises of a better tomorrow.

Beyond just reducing emissions, the pledge suggests that AI itself should be used as a tool to solve climate problems. The goal is to create a framework where the technology helps mitigate the damage it helps cause. By setting terms for energy consumption and transparency, the UN hopes to manage the environmental footprint of the AI revolution more effectively.

AI as a climate solution

While the physical footprint of data centers is a concern, many experts believe the computing power of AI offers a path toward solving climate issues. Some advocates argue that AI can find novel ways to be more efficient with materials and energy. For example, some predictions suggest that advanced AI could help reduce greenhouse gas emissions by 2030.

Supporters of the technology, such as the National Artificial Intelligence Association, argue that the fears regarding energy use might be overblown. They suggest that AI will unlock unknown advances in science and technology that will eventually address climate change and improve human lives. They view the technology as a way to solve the very energy usage issues that data centers currently face.

However, some advocacy groups remain skeptical of these promises. They argue that tech companies are selling hope while hiding the actual harms of their operations. This tension between the potential benefits of AI and the immediate reality of its pollution remains a central theme in international climate discussions.

Capturing waste heat

One practical way to reduce the environmental impact of data centers is by capturing the waste heat they generate. Because computing equipment gets very hot, that heat is usually vented into the atmosphere. Companies like Danfoss suggest that this waste heat is the world's largest untapped energy source and could replace natural gas as a heat source for nearby buildings.

The International Energy Agency estimates that data centers could provide enough heat for around 10% of European homes by 2030 if integrated into district heating systems. This approach is already being tested in various locations. For instance, a campus in Texas plans to use recaptured heat to generate additional electricity on-site, while a project in North Dakota aims to feed heat directly into the community.

Experts note that modern data centers are well-suited for this because they often use liquid cooling instead of air cooling. This method provides a constant source of heat at a steady temperature, which is ideal for thermal energy networks. By turning a waste product into a community resource, data centers could potentially become more integrated into local infrastructure while reducing their overall carbon footprint.

What to watch

  • Whether the Antalya AI Pledge leads to binding international standards for data center transparency.
  • The scalability of waste-heat capture systems in different climates.

Sources and links

First reported by The Economic Times on MSN. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news