The true cost of memory-hogging, energy-hogging AI systems isn’t easy to see. We can start to see the consequences of Big Tech racing to push out more AI with the latest sustainability reports from Google and Amazon.
Their reports show the state of their carbon commitments and aggressively offsetting their carbon footprints. Amazon’s carbon emissions have increased by 16% and Google’s by 25% since last year.
Google and Amazon, without naming AI specifically, show evidence that significant and costly adjustments will be required to their corporate functions to achieve their net zero emissions targets.
It’s all about the AI
Google and Amazon acknowledge whatever the cost, their energy use significantly increased since the adoption of AI systems, and continues to increase.
They also discuss carbon intensity which measures the pollution generated by their businesses relative to revenue. China’s used a similar measure during its climate discussions, and during times of rapid increases to its overall carbon emissions.
Both reports spend many pages justifying the use of AI and how it contributes positively to the environment. Like the Bard would say it is a case of ‘the lady doth protest too much’.
It is important to examine the data for a more extensive understanding.
Amazon and Google are actually in a positive place compared to other companies in terms of emissions related to energy consumption. Other companies’ emissions have increased thanks to a lack of investment in renewable energy. This could become a problem for Amazon and Google too, as more investments are being made in natural gas-powered energy plants as the demand for AI is outpacing what renewable energy can provide.
The bigger problem is Amazon and Google’s Scope 3 emissions. Most of Amazon’s and Google’s emissions is Scope 3 consumption. This extensive category includes emissions from the goods and services companies buy and sell and the related pollution from operations. This solves the problem of companies and consumers.
For Amazon and Google, it can be everything from GPUs to the power customers use when operating phones and tablets.
Google reports Scope 3 emissions in two combined categories, which are capital goods and the products sold and their use. Google reports emissions from the use of its products are so small they are not considered material. That is to be expected, given that small electronic devices comprise a large part of Google’s hardware.
Google’s data center expansion remains a problem, along with everything else.
Last year, Google’s Scope 3 emissions increased by roughly 2.1 million metric tons, which is approximately double the emissions for the year 2019, which is the baseline year Google used in its progress evaluation.
Amazon’s escalating Scope 3 emissions components include capital goods, fuel, and energy. Capital expenditures can incorporate both data centers and warehouses. This could be the reason Amazon’s emissions growth has been greater than Google’s. However, it is likely data centers account for a large proportion of the emissions.
“To meet strong customer demand, in 2025 we added more data center capacity globally than any other company, including more than 1.2 gigawatt (GW) in Q4 alone,” Amazon noted in its report.
Hitting a Wall
This is such a large level of investment, and it helps explain why reducing emissions has become significantly more difficult.
For a large number of years, a large portion of Big Tech’s carbon footprint came from the energy used to power corporate offices and relatively small data centers. Most of that impact could be countered by purchasing renewable electricity.
AI has disrupted that strategy.
Renewable energy coupled with battery storage could be used to power a greater portion of the data centers. Instead, some corporations are using fossil fuels to support the dramatically increasing demand for electricity. That will make achieving net zero even more difficult, but it is a reversible trend.
The more difficult challenge will likely be the emissions associated with the construction and furnishing of data centers. Both steel and cement are high-emission construction materials. Some existing and many new manufacturers are working on lower-carbon construction materials. At the moment, none of those are available to Big Tech at the scale Big Tech requires.
Let’s not forget the semiconductors that are driving the AI boom.
Chip factories are energy hungry and most advanced factories are in Asia where energy sourcing is largely fossil fuel based. Many manufacturing chemicals are also greenhouse gases and some are trillions of times worse for global warming than carbon dioxide. The rapid scaling of chip manufacturing has caused Amazon and Google to violate their carbon neutrality.
These problems can be addressed individually. Amazon and Google and other tech companies will continue to be challenged by these issues.
The road to net-zero will cost a lot of renewable energy, investment in the production of cleaner steel and cement, and the likely purchase of untold carbon removal credits.
Although the deadlines are getting closer, the rapid scaling of AI for chip manufacturing has also caused the deadlines to be further away.
