WIND

Will artificial intelligence delay the energy transition?

Companies in the wind sector have been using artificial intelligence to improve the profitability of their projects. However, the demands that AI will place on the energy system as it expands could make the global transition harder to achieve.

RICHARD HEAP

February 14, 2024

  • AI will require more power than some developed nations by 2027
  • Growing demand is set to increase need for wind and solar farms
  • But DNV warns that displacing fossil fuels has “not truly started”

 

Fierce winds roar with might,

Blades turn, power awakens,

Lights kindle, Earth breathes.

 

Is there anything Chat GPT can’t do? Admittedly, I had to give it four tries to produce the “haiku about the power of wind farms” above, and even then I think the last line is poor. Why is Earth breathing? But the rising influence of generative artificial intelligence (AI) tools like Chat GPT is undeniable.

Launched in November 2022, Chat GPT is an AI-driven language processing tool that has gained huge media attention and is, apparently, set to do people like me out of a job because it can answer questions and respond to writing prompts. It is also just one example of generative AI: as well as Chat GPT rivals, other AI tools can create data or images like the distorted otherworldly wind turbines we increasingly see on LinkedIn.

But generating content is just one use of AI. Digital specialists in the wind sector have spent years touting the benefits of AI-based asset management programmes, with promises that they can help wind farm operators to boost returns at projects by being able to predict when parts will fail. This enables operators plan project maintenance work in a way that reduces costly unplanned downtime. AI can also help project owners and grid operators to predict electricity demand and boost the resilience of transmission networks.

However, the increased use of AI represents a challenge for the energy sector too, and could delay global efforts to move to low-carbon power sources.

Power hungry

Data centres currently account for around 1%-1.5% of global electricity use, according to the International Energy Agency. But the IEA has also argued that demands on data centres will grow rapidly as use of AI increases. Compared to traditional forms of data processing, AI uses more energy in two ways: first, it takes significant processing power to train an AI model, with the IEA saying training a single model can use more electricity than 100 US homes use in a year; and second, it requires more energy to provide answers using AI than it would with, for example, a conventional internet search.

Alex de Vries, founder of research group Digiconomist and PhD candidate at VU Amsterdam School of Business & Economics, produced a study in the scientific journal Joule in October 2023 that said AI-related activities globally could exceed the annual power output of countries such as Argentina, the Netherlands and Sweden by 2027. He added that if Google was to change its search engine to an AI system then the power needed for its nearly nine billion daily searches would be equivalent to the electricity consumption of Ireland.

The growth of AI is therefore a challenge and opportunity for wind companies.

We heard earlier about some of the uses of AI in the wind sector, and they will keep growing in the coming years. The theory that AI can help firms to keep driving down the cost of electricity from wind farms is long established. We can still remember turbine makers singling out the potential of AI to keep reducing project costs at our annual conference in 2019, but with little detail about how that would be done.

Growing demand for electricity presents an opportunity for the continued expansion of wind and solar globally. These technologies have been adept at meeting new electricity demand around the world in recent years, and we see evidence that countries are seeking to improve permitting to enable both wind and solar to keep playing this role. Last week, WindEurope said that national governments are making progress to improve permitting for new projects, for example. Wind farms will be benefit from increased demand for green energy.

And yet, we must also be wary of what this means for the energy transition. It is all very well for wind and solar to meet new demand for electricity around the world. These are cost-effective forms of new electricity generation. But if the main role of this new renewables capacity is to cater for new power demand, that means we are not using them to replace incumbent fossil fuel generation.

This is a concern that pre-dates any forthcoming explosion of AI. In its ‘Energy Transition Outlook 2023’ in October, DNV said the energy transition has “not truly started” if we define a ‘transition’ as when “clean energy replaces fossil fuel energy in absolute terms”. It predicted that wind and solar would increase 13-fold by 2050, but did not clarify how much would replace fossil fuels.

“Up to the present, renewables have met some, but not all, of the world’s additional energy demand,” it said, adding that the energy transition “seems to be in stall mode”.

This is the crux of the matter as AI continues its rapid expansion. It has huge potential to re-shape all facets of society, the energy system included, but we must be aware of how this could stall the transition too. But one thing is for certain: the sheer power that AI will need as it becomes a bigger part of our daily lives will be too large for the wind industry to ignore.