‘Artificial intelligence will make it easier to combat climate change.’ That’s the bold claim made by billionaire and self-proclaimed climate problem–solver Bill Gates. But Gates – a well-known advocate of geoengineering – is not alone in promoting the idea that technology can be our saviour. Seden Anlar reports.

Jeff Bezos, another billionaire and the CEO of Amazon, one of the world’s most polluting companies, has also placed his bets on AI to address climate change. Through the Bezos Earth Fund, he has already committed EUR 94 million in grants for AI solutions aimed at tackling climate change and curbing biodiversity loss.

Over the past few years, AI’s role in addressing the climate crisis has gained such momentum that, for the first time, it was a topic of discussion at last year’s COP28 climate summit in Dubai.

It is true that AI could be used to track deforestation, identify pollution leaks and track extreme weather events. In fact, AI-driven technologies are already predicting droughts in Kenya, measuring changes in melting icebergs, mapping the carbon absorption capacity of forests in Indonesia and providing communities in Malawi with flood warnings 15 days in advance, which helps them evacuate in time. Similarly, AI-driven drones in Brazil are being guided by AI-powered computers to reforest the hills around Rio de Janeiro, carefully planning how many seeds should be planted and where.

In Europe, too, AI-driven technologies are being used to tackle environmental issues. For example, an environmental organisation called The Ocean Cleanup in the Netherlands is using AI technology to clear plastic pollution from the ocean.

At what cost?

While the hype around AI centres on the promise that it will help save the planet, the reality is that AI-driven technology and the industry surrounding it are currently doing just the opposite.

That’s because the complex algorithms and machine learning models powering AI aren’t powered by thin air; they require enormous computational power from thousands of servers housed in data centres around the world. These data centres, operated by the tech giants pushing AI, consume vast amounts of electricity and water to keep running.

What’s even more alarming is not just that data centres require energy but the sheer amount of energy they consume. According to researcher Jesse Dodge, ‘one query to ChatGPT uses approximately as much electricity as could light one light bulb for about 20 minutes’. Multiply that by millions of users daily, and the total energy consumption becomes staggeringly high. So this isn’t just a minor hiccup in our fight against climate change – it’s a glaring contradiction.

To put things into perspective, there are over 8,000 data centres worldwide, a number that has nearly doubled since 2015. Collectively, these centres now consume as much electricity as the entire country of Italy. And it doesn’t stop there. As AI becomes more widespread and AI tools grow more sophisticated, energy demand will only increase. In just three years, AI servers could consume as much energy as all of Sweden or the Netherlands, according to research.

Tech companies are fully aware of this growing energy appetite and are responding by building even more data centres. For instance, Microsoft plans to spend USD 50 billion on expanding its data centres within a single year, while Google and Amazon continue to invest billions in new data centre projects across the globe. The construction of these data centres isn’t environmentally friendly either; it requires carbon-intensive materials like steel, cement and semiconductors, further contributing to the climate crisis they claim to address.

But it’s not just about the volume of energy consumed; it’s also about the source of this energy. As AI becomes more sophisticated and widespread, the demand for power intensifies, and much of that power, for instance in the US, still comes from burning fossil fuels like coal and gas – the very culprits behind climate change.

Pledges versus reality

This rapid increase in energy usage and demand is already impacting the climate pledges that major tech companies set for themselves before the AI hype cycle.

Companies like Google and Microsoft had pledged to achieve ambitious climate goals: Google aimed to operate on carbon-free energy 24/7 by 2030, while Microsoft promised to be carbon-negative by the same year. However, as they continue to pour billions into AI development, these targets are slipping further away.

Microsoft’s latest sustainability report revealed that its emissions have risen by 29% since 2020, attributing this spike directly to the expansion of data centres optimised for AI workloads. Google also admitted that its greenhouse gas emissions surged by 48% over the past five years, largely due to the increased energy demand from AI deployment. Despite its promises of reaching net-zero emissions by 2030, Google has already backtracked on its claim of ‘maintaining operational carbon neutrality’, prioritising its drive to build more advanced AI systems at any cost.

These tech giants claim they’re working on alternative energy solutions. Google boasts that it’s using machine learning from its AI lab DeepMind to reduce energy usage for cooling data centres by 40%. But these efforts seem like mere drops in the ocean, as the majority of data centres are still powered by fossil fuels, putting us further behind the climate targets that need to be reached to maintain global temperature rise within manageable limits. With this level of energy demand, even when it seems like these tech giants are reducing their carbon footprint, it’s often due to carbon offsets they purchase to balance out their emissions. And even when they do tap into renewable energy, they’re often just buying power from existing renewable power plants, thus diverting these resources away from other sectors that desperately need them, instead of investing in additional renewable power generation capacities.

Europe’s dilemma

The situation is no different in Europe. The International Energy Agency (IEA) states in its 2024 report that of the 8,000 data centres worldwide, a third are in the United States and 16% are in Europe.

As Europe’s AI ambitions expand, so too does the demand for data centres and the fossil fuels needed to power them. In just one year, from 2023 to 2024, Europe’s market saw a 168% increase in data centre investment, according to research by the London-based global real estate consultancy Knight Frank.

One European country significantly affected by energy-hungry data centres is Ireland. Major tech companies like Google, Meta, Amazon and TikTok already run over 80 data centres in the country, with several more expansions and new facilities planned. In fact, Ireland’s data centres consumed more electricity in 2023 than all of its urban homes combined, according to official figures.

Over the past few years, other European cities such as London, Paris and Frankfurt have also emerged as data hub hotspots. Amazon’s cloud computing division AWS announced plans to invest EUR 9.5 billion over the next five years to build and operate data centres in the UK, while, similarly, Microsoft is investing EUR 4 billion in French data centres.

With these and many more expansions to come, the European Union can expect data centre electricity consumption to rise from just under 100 terawatt-hours in 2022 to almost 150 terawatt-hours by 2026, according to the IEA. Moreover, the EU estimates that by 2039, data centres in the bloc will consume more than 3% of all energy used in the EU if this trend continues.

Will the EU take action?

In terms of how the EU will respond to this trend, the previous mandate primarily focused on regulating AI technologies, which included introducing an EU-wide scheme to assess the sustainability of data centres. As a result, data centres are now required to file reports detailing their energy and water consumption, as well as the steps they’re taking to reduce it.

However, it appears that in the next mandate, the EU’s focus will shift towards boosting the European tech industry, aiming to make the EU more competitive in tech and AI on a global scale. The European Commission has already started moving forward with efforts to facilitate AI data centres, including the launch of an initiative to establish ‘AI Factories’ – large-scale infrastructure designed to support European AI startups and small-to-medium enterprises. Through these AI factories, the EU aims to link AI startups with its EU-funded network of supercomputers to boost innovation and reduce dependency on US computing power.

This AI Factories Initiative is part of a broader package the Commission put forward in January 2024, and by early September, it had already issued a call to kickstart the project.

However, the EU might be late to the party, as many of Europe’s most promising AI companies have already partnered with the US Big Tech players, to use their access to computing power. For example, France’s AI startup Mistral reached a deal with Microsoft in February to train its technology on Microsoft’s supercomputers.

The bottom line

Isn’t it ironic that we’re banking on a technology we don’t have the resources to power sustainably, expecting it to save us from a problem it’s actively contributing to?

The figures may seem overwhelming, and AI might be touted as the solution to the climate crisis, but in reality, it appears to be taking us further down the path of destruction.

This is why it’s time to question whether AI is truly the answer or just another shiny distraction – another false saviour – diverting us from the urgent need to address the root causes of climate change: our relentless pursuit of growth, consumption and profit at the expense of the planet.

 

The views and opinions in this article do not necessarily reflect those of the Heinrich-Böll-Stiftung European Union.