It’s well known that AI is a resource guzzler, utilizing energy and water unlike any digital technology prior. Now, the toll on the environment due to Big Tech’s AI ambitions is coming to light.

This week, both Google and Amazon released their sustainability reports, revealing troubling figures. Each company aims to achieve net-zero carbon emissions in the near future, yet the demands of AI are complicating those objectives. Google has seen its total carbon emissions rise by 25% from last year, while Amazon’s have increased by 16%.

A thorough examination of the reports indicates that both Amazon and Google will need to implement significant, and potentially expensive, changes to their operations in order to meet their net-zero ambitions.

Neither firm explicitly attributes rising emissions to AI, but there is substantial indirect evidence suggesting this link.

AI at the core

Both Amazon and Google acknowledge a substantial rise in their energy consumption over the past year, coinciding with increased AI adoption. They reference carbon intensity — essentially, the pollution produced per dollar of revenue — a metric that China has leveraged during climate negotiations amidst its soaring emissions. Moreover, both companies spend multiple pages highlighting the potential environmental benefits of AI, which could be seen as an overreaction to scrutiny.

The situation becomes clearer as you delve deeper into the details. Both companies are performing relatively well in terms of carbon pollution associated with energy procurement. Years of investing in renewable energy have prevented significant increases, although this may shift soon as tech firms, including Google, are now heavily investing in natural gas power plants to meet the demands of AI.

Most of the increasing carbon emissions for Amazon and Google stem from Scope 3 emissions — a broad category that includes pollution from activities outside a company’s direct control, such as the goods and services purchased or the products sold. For companies like Amazon and Google, Scope 3 encompasses GPU acquisitions and the usage of their products, like smartphones and tablets.

Google combines two types of Scope 3 emissions — capital goods and the use of sold products — even though it acknowledges that the latter is minor enough to be immaterial. (The majority of Google’s hardware products are compact devices that don’t consume substantial energy.) This likely positions data centers as the primary contributor. Last year, Google’s Scope 3 emissions rose by 2.1 million metric tons, meaning they have now doubled since 2019, which serves as the baseline year for assessing performance.

Amazon’s increasing Scope 3 emissions primarily arise from capital goods as well as fuel and energy. The former might include data centers and warehouses, providing insight into why Amazon’s Scope 3 emissions have surged more significantly than Google’s. Nevertheless, a considerable portion is likely also attributed to data centers. “To satisfy vigorous customer demand, in 2025 we expanded global data center capacity more than any other company, including over 1.2 gigawatts (GW) in Q4 alone,” Amazon stated in the report.

Hitting a barrier

Such expenditures shed light on why achieving decarbonization has suddenly become much more challenging. Previously, energy for offices and moderately sized data centers was the biggest contributor to their carbon footprints, which could have been offset by purchasing renewable power. 

AI has disrupted that paradigm. While tech companies could continue utilizing renewables alongside batteries to power their data centers, they are increasingly reverting to fossil fuels. This trend complicates the fulfillment of their net-zero commitments, although it is not irreversible.

The more troublesome emissions arise from constructing and equipping data centers. The steel and cement sectors are significant polluters, and while startups are exploring low-to-zero carbon solutions, they are not yet prepared to operate at the scale required by tech firms. 

Additionally, there are the GPUs and memory chips fueling the AI surge. Semiconductor fabrication requires immense energy, and many top-tier chip manufacturing plants are situated in Asia, where electricity grids largely rely on fossil fuels. Compounding the issue, many of the chemicals used in these factories are also potent greenhouse gases, capable of contributing to atmospheric warming at rates thousands of times greater than equivalent CO₂ emissions. The surge in chip consumption has likely exacerbated both Amazon’s and Google’s carbon footprints.

However, none of these challenges are insurmountable; yet Amazon, Google, and their contemporaries face a significant undertaking. To fulfill their net-zero pledges, they will need to significantly increase their renewable energy acquisitions, invest substantially in advanced steel and cement production methods, and procure millions of tons of carbon removal credits. Achieving this remains feasible, but their commitment to AI hasn’t simplified matters.