AI is Solving Climate but is Leading us to a Catastrophe.

AI is revolutionising many industries, including the climate industry, but at the cost of the planet. Unimaginable terawatt hours of electricity and billions of gallons of water are guzzled to cool server farms housed in numerous data centres.

The International Energy Agency (IEA) released a forecast report about the global energy required for our electricity consumption until 2026 (especially with the surge in AI use, data centres, and crypto). GenAI needs more data centres to run and cool the servers, and this overconsumption is stressing our planet.

What does climate change have to do with it?

Nvidia is leading the AI data server market, and GenAI already matches the energy needs of a small nation (like Lithuania) that could power about 4 million people! Many reports have mentioned the ridiculous amounts of energy GenAI models (especially for image generation) need to get trained in infinite parameters. The computational power required to train the models doubles every nine months and is not even slowing down—and if they outpace the power generation, we will see brownouts and blackouts.

The concern is that energy needs may double in 2026 (as much as Japan would consume). Also, consuming such electricity (the fossil-fuel-based one) produces more carbon emissions, which would take several trees to work for a year to absorb. By 2028, it could use more energy than Iceland used in 2021! By 2030, data centres will use a quarter of American electricity.

Alphabet, Meta, Open AI, Microsoft, Amazon, Nvidia and even Silicon Valley are now infamous for their high GHG emissions due to their employment of AI systems and their non-transparency in their carbon footprint.

The need for AI is undeniable, but can it go green?

Currently, data centres are being made more sustainable by being powered by renewable sources—Microsoft is eyeing Hydrogen fuel cell power for that, and Amazon owns nuclear-powered data centres. However, ironically, many countries use smart grids for renewable energy efficiency and optimisation!

Microsoft did a trial run with underwater data centres but is closing its experimentation. The company has released statements that it wouldn’t be setting up data centres undersea anymore but will use the learnings to simulate the conditions of such environments for data centres to be more sustainable and effective and with zero human intervention for maintenance (yes, they have bots for that).

I can also dive deep into how that might have wrecked the ocean ecosystem in a separate piece.

While dealing with this cluster chaos, other initiatives to develop innovative and sustainable cooling systems for data centres are underway. Developing energy-efficient AI chips or AI algorithms would help. Consumers need to use energy-efficient devices. Also, using shared data centres and cloud computing could lower the footprint for many instead of commissioning private infrastructure.

We will cover climate solutions for AI later.

The surge in GenAI will spike renewable energy demand.

Bill Gates has stated that GenAI will find ways to make data centres more efficient and sustainable. He says that GenAI will fix itself eventually as it will find ways to undo the damage that’s been done to train it.

However, GenAI’s boom will demand more energy, consequently requiring more renewable energy. This requirement would lead us to a faster transition from fossil fuels to renewable energy. Can we call this a win? No. AI’s demands are outpacing renewable energy production, pushing our economies into an energy crisis catastrophe.

Our older articles showed how building renewable energy capacities already causes environmental concerns. Every piece of equipment carries embodied carbon—can we conveniently forget that?

Check out these articles: Dark Side of Renewable Energy and the Geopolitics of Decarbonisation

I am not advocating for degrowth or disregarding AI’s intervention, which could bring novel solutions to fight climate change. We are now highly dependent on it (I am too)—be it carbon capture storage tech, weather prediction breakthroughs in nuclear fusion, energy optimisation/storage solutions in green technology, or helping research and develop new materials for clean energy—AI has a pivotal role.

Interestingly, Sam Altman owns both OpenAI and Helion (a nuclear fusion company), so I will discuss energy politics in the upcoming articles.

We ought to collectively worry about how AI is helping us evolve into a climate-smart civilisation but will disturbingly be one of the leading causes of pollution.

International policies must be robust. We could also put a carbon tax on AI companies, but only if we know the level of emissions we are dealing with.

The top dogs of the AI industry are refusing their disclosures and playing power games to monopolise the non-renewable and renewable energy market (especially nuclear) to feed their servers. It is high time world governance kicked in, or we will lead ourselves to irreversible consequences where we may destroy ourselves before climate change does.