Could AI's Energy Demands Harm Google's Environmental Goals?

AI is rapidly reshaping industries, offering advancements in efficiency and technology, yet it also introduces new challenges regarding sustainability.

Google’s report titled "The AI Opportunity for Europe’s Climate Goals - a Policy Roadmap" explores how this tech can be supported to become a force for good.

While renowned for its environmental commitments, the tech giant faces scrutiny for the expanding environmental footprint of its AI initiatives.

AI’s environmental paradox

Google has long positioned itself as a pioneer in corporate sustainability. 

It became the first major company to reach carbon neutrality in 2007, has matched 100% of its electricity consumption with renewable energy since 2017 and is aiming to operate entirely on carbon-free energy (CFE) by 2030.

Training and running large-scale AI models such as Gemini, Google’s flagship generative AI system, is a key aspect of Google’s path towards environmental protection.

This system is integrated in a variety of Google products, including Google Earth, Docs and Gmail. 

“Google Earth has democratised geospatial information for a wide range of users and use cases. It renders a 3D representation of Earth, allowing people to explore our planet from endless vantage points,” explains Kate Brandt, Chief Sustainability Officer at Google.

“Businesses utilise its layers to analyse potential renewable energy sites and with new Gemini capabilities, they can ask questions, analyse land data and make informed decisions faster when siting a solar farm.”

However, Gemini requires immense computing power. 

These workloads run across Google’s vast cloud infrastructure, adding significant pressure to already energy-hungry data centres.

“Google’s AI efforts are built on the same infrastructure as its core services,” notes the report, including YouTube, Search and Gmail.

Unlike traditional services, AI models must be trained on huge datasets and require constant updates – intensifying both energy use and water consumption.

Rising emissions vs climate commitments

By 2023, emissions from Google's data centres increased by 48% compared to 2019 levels.

While this is partly due to expanded operations and increased cloud usage, many experts point to AI workloads as a key factor in this rise.

This presents a dilemma for Google, which often celebrates its achievements in clean energy and environmental innovation, yet faces criticism for the growth induced by AI.

“Google’s AI ambitions are incompatible with its climate goals,” says Fieke Jansen, a researcher at the University of Amsterdam’s DATACTIVE project. 

“You cannot continue to grow your emissions and claim to be on a path to sustainability.”

The transparency gap

There is significant concern over the lack of detailed transparency regarding the environmental impact of AI systems.

While Google provides annual sustainability reports, specifics on energy and water use linked to individual products, primarily its AI, remain undisclosed.

In February 2024, the company announced it would embed its Gemini model across Gmail, Google Docs and Android smartphones – vastly expanding AI’s reach and, potentially, its carbon footprint.

“Without disaggregated reporting, there’s no way to verify the impact of AI specifically,” says Fieke. 

“That makes accountability nearly impossible.”

This opacity is not unique to Google.

Across the tech sector, AI's environmental reporting is often generalised, hindering efforts to thoroughly evaluate its ecological impact.

Water scarcity and local impact

Besides energy, water use is a pressing issue, as data centres rely on evaporative cooling systems that require substantial water volumes.

In 2021, Google's centres consumed 5.6 billion gallons globally.

This figure is likely to have increased in the years since as AI models and infrastructure are scaled up. 

In regions like The Dalles, Oregon, where Google operates a large data centre, local communities have raised alarms about the pressure on limited water resources, especially during periods of drought.

The combination of high energy use, water dependency and increased heat generation raises significant sustainability challenges – not just globally, but locally.

Sustainable AI

In response to mounting scrutiny, there have been some positive signals from Google and others in the industry. 

Google DeepMind’s 2023 initiatives aim at enhancing model efficiency via refined architectures, reducing redundancies and smarter scheduling to lower energy usage.

Researchers call this “green AI,” though much remains in early phases relative to AI deployment's swift pace.

“There is no doubt that AI has enormous potential to help solve global sustainability challenges,” says Fieke. 

“But the way it’s currently being scaled raises serious questions about whether the solution is becoming part of the problem.”

Microsoft, Meta and Amazon are all racing to integrate generative AI into their platforms and services, while simultaneously publishing ambitious climate pledges. 

The result is a growing disconnect between digital expansion and environmental stewardship.

As more governments introduce legislation mandating environmental disclosures and lifecycle assessments for digital technologies, tech giants may face increasing pressure to demonstrate not just ambition, but accountability.

Explore the latest edition of Energy Digital Magazine and be part of the conversation at our global conference series, Sustainability LIVE.

Discover all our upcoming events and secure your tickets today.

Energy Digital is a BizClik brand