Google and Amazon's carbon emissions surged in 2025, with the AI boom outpacing the grid's ability to transition to clean energy. Google reported a 16% annual jump in emissions, while Amazon saw an 18% rise compared to 2024. Over longer timescales, the picture is even starker: Amazon's emissions have climbed 58% since 2019, and Google's have jumped 82% in the same period.

The timing of these disclosures coincides with a massive US heatwave that is testing the nation's power infrastructure and raising urgent questions about whether the country can support the rapid expansion of AI data centers without compromising grid stability and water security.

Why Are Tech Giants' Emissions Growing So Quickly?

Both companies attribute their rising carbon footprint to the same culprit: the construction and operation of AI data centers. These facilities consume enormous amounts of electricity to train and run artificial intelligence models, as well as to cool the servers that power them. Google's chief sustainability officer, Kate Brandt, acknowledged the tension directly.

"Our AI infrastructure buildout is currently accelerating faster than the grid is decarbonizing," said Kate Brandt.

Kate Brandt, Chief Sustainability Officer at Google

Amazon's electricity consumption from purchased power rose by 34% in the last year alone, driven largely by data center expansion and the electrification of its delivery network. Google pointed to the resources needed to construct and operate data centers, plus the supply chain for chips and servers, as key drivers of its growing emissions.

To put these numbers in perspective, Amazon's 80.85 million tons of CO2-equivalent emissions last year exceed the entire annual carbon output of Austria or Greece. Google's 18.8 million tons, while smaller in absolute terms, represents a dramatic acceleration. A UN report published in June 2026 found that if data centers were a country, they would rank 11th globally in electricity consumption, sitting between France and Saudi Arabia.

How Are Heatwaves Exposing Data Center Vulnerabilities?

The current US heatwave is creating real-time stress on power grids already stretched by data center demand. On the East Coast, PJM Interconnection, the nation's largest power grid operator, asked the US Department of Energy to order data centers to switch to backup power generators within 15 minutes during emergencies. This request would free up power for residential and commercial customers during peak demand periods.

The challenge is particularly acute because data center cooling systems consume as much as 40% of a facility's electricity during normal temperatures, but that percentage climbs significantly as outdoor temperatures rise. During extreme heat, cooling demands can spike dramatically, precisely when the grid is already strained by residential and commercial air conditioning use.

Data centers are also creating localized heat islands around their facilities. A University of Cambridge study found that temperatures increased by an average of 2 degrees Celsius near data centers, with some areas experiencing increases as high as 9.2 degrees Celsius in the immediate vicinity.

What Are the Water Implications?

Beyond electricity, data centers consume vast quantities of water for cooling. A single large AI data center can use as much as 5 million gallons of water per day, with consumption typically spiking during the hottest periods of the year, when many communities are already facing water shortages.

The problem is compounded by geography. Two-thirds of all new data centers built or in development since 2022 have been located in water-stressed regions. Many cooling systems do not recirculate water, meaning nearly 80% of the potable drinking water used in data center cooling simply evaporates.

Currently, data center water usage stands at 627 million gallons per day, which is comparatively modest compared to agriculture and residential use. However, as AI demand surges, so will data center water consumption, creating conflicts in regions already managing scarcity.

How Are Policymakers and the Public Responding?

Public opposition to data center expansion is growing. According to a recent Gallup survey, seven in 10 Americans oppose data center construction in their local communities, with half citing excessive resource consumption as their primary concern.

Across the political spectrum, lawmakers are calling for action. Texas Governor Greg Abbott called for a ban on building data centers in rural areas, while also suggesting that data centers should generate their own power and reuse water. On the left, Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez have called for a moratorium on new data center construction.

The urgency is underscored by projections from the US Department of Energy. Data centers currently account for 4% of US power demand, but that figure is expected to double to 9% by 2030. A typical hyperscale data center requires between 100 and 300 megawatts of electricity, enough to power as many as 300,000 US homes or a city of roughly 750,000 people.

Steps Utilities and Companies Are Taking to Address the Crisis

• Backup Power Requirements: Grid operators like PJM are requesting that data centers maintain backup power generators and switch to them within 15 minutes of emergency signals, reducing strain on the main grid during peak demand periods.
• Efficiency Claims: Both Google and Amazon are emphasizing their efficiency improvements. Google claims that without its decarbonization initiatives, 2025 emissions would be five times higher. Amazon states its data centers are more water and energy-efficient than the industry average.
• Infrastructure Investment: Utilities across the country are accelerating investment in new generation and transmission lines while delaying the retirement of aging power plants to meet growing demand from data centers, electrified transportation, and new manufacturing.

The challenge facing both tech companies is that their climate pledges are now in jeopardy. Amazon has committed to reaching net-zero carbon emissions by 2040, but acknowledged in its latest report that AI technology may slow progress toward that goal. Google's pledge to cut 2019-level emissions in half by 2030 faces similar headwinds.

"We may be able to move faster, or the demand may slow us down," acknowledged Kara Hurst.

Kara Hurst, Chief Sustainability Officer at Amazon

The convergence of record emissions growth, grid strain during extreme weather, and public opposition suggests that the AI boom's environmental costs are no longer abstract concerns. They are immediate, measurable pressures on the infrastructure that powers modern life. As AI demand continues to surge, the question facing policymakers, utilities, and tech companies is whether the grid and water systems can adapt fast enough, or whether the pace of data center expansion will need to slow.