[Image of Google, Credit to Pixabay]

Google’s carbon emissions surged nearly 50% compared to 2019, a significant increase driven by the expanded use of AI.

Google’s emissions rose by 13% over the year 2023, producing 14.3 million metric tons of CO2 pollution, which is equivalent to the amount of CO2 that 38 gas-fired power plants might release annually.

Due to the sudden increase in the carbon emission rate, it became increasingly difficult to achieve its initial goal of cutting its planet-heating pollution in half by 2030 compared to a 2019 baseline.

For the increased emission rate, the two most significant contributing factors were the rapid advancements in and demand for AI.

AI models require substantial amounts of energy since they perform a huge number of calculations in short order.

Particularly since Google has fallen into the Gemini era, redesigning Search with generative AI and injecting AI into its other products have greatly contributed to the increased emission rate.

In fact, Google admitted the challenge of reducing the emission rate, saying “As we further integrate AI into our products, reducing emissions may be challenging due to increasing energy demands from the greater intensity of AI computation and the emissions associated with the expected increases in our technical infrastructure investment.”

According to the International Energy Agency (IEA), the world’s data centers use around 1 percent of the world's electricity.

However due to the boom in the AI industry, the IEA recently estimated that it could consume 10times as much electricity in 2026 as it did last year.

In other words, AI data centers are expected to add about 323 terawatt hours of electricity demand in the U.S.

In reality, electricity consumption, mostly from data centers, contributed to nearly a million metric tons of pollution to the company’s carbon footprint in 2023 and made up the biggest source of Google’s additional emissions last year.

Renewables that will ameliorate this situation might emerge, yet immediate implementation is challenging.

Despite the challenge of immediate implementation, Google strives to minimize its environmental impact by making more energy-efficient AI models, hardware and data centers .

More specifically, Google boosts the training efficiency of its fifth-generation TPU by 39% with techniques that accelerate training, like quantization, where the precision of numbers used to represent the model’s parameters is reduced to decrease the computational load.

Moreover, it also deployed carbon-intelligent computing platforms and demand response capabilities at its data centers.

Furthermore, Google has also implemented a water stewardship program, which aims to minimize water consumption and reduce waste, offsetting 18% of its water usage in 2023.

However, Google is not the only tech giant grappling with increased emissions due to AI demand.

Microsoft’s total carbon emissions rose nearly 30% since 2020, attributed to the construction of data centers.

Amazon, which hasn’t disclosed its 2022–2023 figures, said that between 2021 and 2022, emissions decreased slightly by 0.4 percent, though they were still up almost 40 percent overall since 2019.

"The growth trend is super-fast," Fengqi You, an energy engineering professor at Cornell University, said. He further added, "This is something I'm concerned about."