AI has been touted as the solution to the energy transition, but its massive energy requirements appear to sit in contrast with its potential.
The technology promises to streamline asset maintenance processes, efficiently manage intermittent renewables and balance supply and demand pressures on the grid. However, the massive and expanding data centres required to power the rollout of AI are energy-intensive, and threaten to unpick the technology’s positive potential.
How well do you really know your competitors?
Access the most comprehensive Company Profiles on the market, powered by GlobalData. Save hours of research. Gain competitive edge.
Thank you!
Your download email will arrive shortly
Not ready to buy yet? Download a free sample
We are confident about the unique quality of our Company Profiles. However, we want you to make the most beneficial decision for your business, so we offer a free sample that you can download by submitting the below form
By GlobalDataEarlier this month, Google shared that its greenhouse gas (GHG) emissions have climbed 48% over the past five years, whilst Microsoft recently revealed that its overall scope 1–3 emissions have increased by 29% compared with 2020, with the rise primarily attributable to the expansion of its data centres.
The International Energy Agency (IEA) reports that AI could use up to 1,000TWh annually by 2026 – equivalent to Japan’s electricity consumption. However, speaking to Power Technology, experts say that the energy-intensive demands of AI could be justified by the security it provides, and its potentially crucial role in managing clean energy offerings.
Andy Pardoe, chair of the Deep Tech Innovation Centre at the University of Warwick, explains that “AI not only offers solutions in optimising energy production, distribution and storage but also results in more efficient and sustainable systems. For example, AI-driven predictive analytics contribute to renewable energy integration and improved grid stability, while enabling tighter demand-response mechanisms. On balance, I think AI’s contributions significantly outweigh any potential hindrances it may present.”
Achieving grid security
Grids are under pressure, and wind and solar projects that promise to expand clean energy capacity can face delays of up to 15 years before connection. In a 2023 report, the IEA noted that, “at least 3,000 gigawatts (GW) of renewable power projects, of which 1,500GW are in advanced stages, are waiting in grid connection queues – equivalent to five times the amount of solar PV and wind capacity added in 2022”.
An efficient grid will be essential then. John Langley-Davis, future network manager at Lucy Electric, tells Power Technology that “AI is already having a significant impact on the UK’s energy grid. AI-enabled grid monitoring is revolutionising how we maintain and operate our energy distribution networks. Imagine being able to know the health status of cables, which are currently invisible assets underground, without digging them up. With this AI-enabled visibility, we can accurately plan asset maintenance and replacement, just as we would with assets like transformers that we can physically inspect.”
It is a point echoed by Taco Engelaar, Europe Lead at Neara, who states: “Some AI-powered technologies are helping us solve the ‘gridlock’ and freeing up capacity for more renewables to be connected. When used alongside digital modelling technologies, AI offers unparalleled visibility into the additional capacity that exists within power lines all over the world. For example, a recent study with EMPACT Engineering found that 94.5% of power lines in a fast-growing region of Texas could safely run at double the current capacity.
“By finding room in existing infrastructure, we can release projects gathering dust in connection queues and speed up the clean energy transition. Used responsibly, there is huge potential to build a reciprocal relationship between AI and energy infrastructure.”
However, he caveats this with an acknowledgement that “there is no getting away from the fact that Generative AI [GenAI] is hugely power-intensive”.
“As grid constraints stall all sorts of worthy projects, from renewables to new-builds, the added strain from GenAI can appear at odds with the clean energy transition – but not all AI is the enemy of climate progress.”
The energy transition needs infrastructure
The unpreparedness of the grid for the energy transition is symptomatic of a greater need for more infrastructure around clean energy. It is a point made by cybersecurity specialist Michael Robert, who says that, on its own, “AI is not a silver bullet”, calling for increased “investments in clean energy infrastructure and strong mechanisms for accountability and transparency”.
Daniel Efrati, CEO of Ned Data Centers, considers the global picture: “There are clear signs that existing power structures globally are not enough to meet the demands of the future, fuelled by AI.
“Some countries have struggled with the strategic planning of their infrastructure and have chronically underinvested in their grids. Such countries will suffer from insufficient power due to the demands of data centres. Without this investment, governments will find themselves losing the services that data centres provide, and relying upon other countries for their internet and data storage needs, which include offshoring sensitive and private data, and maybe even facing power blackouts.”
Efficient and developed infrastructure is an essential for AI to be a force of good, but the IEA reports that electricity consumption from data centres, AI and the cryptocurrency sector could double by 2026.
Bruce Torrance, senior associate at specialist intellectual property law firm Reddie and Grose, tells Power Technology that “for better or worse, the AI genie is out of the bottle… I am not blind to the negatives”.
In addition to the need for more infrastructure around clean energy, he points to further innovation in AI efficiency, to power a technological revolution that is set to stay. “Curbing the growth [of electricity demand] will require technological advancements or drastic actions: large US tech companies are already looking to commission their own nuclear power plants to fulfil their energy requirements at low carbon cost.
“I remain positive. AI may be a problematic emerging technology, but these problems are not insurmountable. Patent publications for quantum computing, which could make AI more efficient, have increased ten-fold in the last decade, and there is increasing confidence that nuclear fusion will provide plentiful clean energy to the grid before the end of the next decade.”
Solving the intermittent nature of renewables
More than 30% of global electricity was generated from renewable sources in 2023, according to a report by global energy think tank Ember, which attributed the increase in generation to growth in the wind and solar sectors.
However, both areas pose a clear issue: they generate power intermittently, depending on environmental conditions.
AI holds promise for stabilising supply and demand, as chief technical officer at Spectrum Search Peter Wood explains. “In enhancing the efficiency of renewable energy sources, AI is truly invaluable. Solar and wind power, whilst plentiful, are unpredictably intermittent. AI algorithms can foresee weather changes and adjust the functioning of these renewable energy sources accordingly, amplifying their output. It determines the best times for storing solar energy or distributing it to the grid, ensuring that every bit of energy generated is put to good use,” he explains.
Muhammad Wakil Shahzad, assistant professor of mechanical and construction engineering at Northumbria University, adds: “AI can predict energy demand more accurately, optimise the operation of power grids and manage the intermittent nature of renewable energy sources like solar and wind. By improving energy storage and distribution, AI can reduce waste and ensure that clean energy is used more effectively. Despite its power requirements, the benefits of AI in creating smarter, more resilient energy systems outweigh the drawbacks. Additionally, advances in AI efficiency and the increasing availability of renewable energy to power data centres can mitigate concerns about AI’s energy consumption.”
Working in the offshore wind sector, CEO of Rovco Brian Allen considers AI’s benefits. “In offshore wind, it is already proving its value. There are many processes across the life cycle stages of an offshore wind farm that can be automated and accelerated by using AI, from counting and measuring boulders while planning a wind farm, to assessing the integrity of blades or foundation structures. Without AI, these time-intensive jobs have to be done manually by offshore workers… This potential of AI to make processes more efficient and improve productivity is all the more vital given the worker shortage the renewables industry is facing.
“In the UK offshore wind sector, estimates from the Offshore Wind Industry Council (OWIC) suggest we need an additional 70,000 workers by 2030. If we don’t have the workforce we need, it will be near impossible to reach the UK’s current target of 50GW capacity by 2030. The use of AI to support and enhance the workforce isn’t just an option, it is a necessity for the success of the offshore wind sector.”
AI: a help or a hindrance?
AI is undoubtedly set to stay, and integrating the technology with solutions to the transition will be a priority for energy companies over the next decade. The emissions associated with the technology pose a problem (training GPT-3 generated 502 tonnes (t) of carbon, and the model now emits an additional 8.4t of CO₂ annually), but the balance appears to be in favour of AI.
Rob Mowat, managing director of ESS Expo, says that “while concerns exist around its power requirements leading to increased energy consumption, the benefits of AI in enhancing grid efficiency, improving battery storage and informing environmental decisions must surely far outweigh the negatives. The key lies in responsible deployment, ensuring AI fosters the transition to clean energy rather than exacerbating existing challenges.”
George Borovas, partner at Hunton Andrews Kurth LLP and board of management member at the World Nuclear Association, recognises that AI is here to stay and agrees that this is likely good news overall. “There is no question that AI will transform every aspect of society including the efficient production and usage of energy,” he says.
“But while in the long term AI may end up reducing overall energy consumption, the development of AI itself will need reliable and abundant power in the near future. Nuclear small modular reactors may become the key power source that will provide the necessary stable and plentiful power 24/7, allowing the AI companies to develop their ambitious plans.”
Chris Sadler, founder of Kimble Solar, is clear that AI is good news for the energy transition: “It will help the transition. There is no stopping AI, so there is no point in wondering if the extra power consumption needed for all those processors will be matched by the energy-saving benefits.
“Just like mining for Bitcoin, servers and processors are guzzling up energy with our everyday digital needs, but AI will 100% improve our transition to clean energy. By using algorithms to predict our energy usage patterns and forecast the availability of renewable energy, AI can efficiently manage the charge and discharge of batteries and perform load-shifting to optimise our energy use. The real question is, is the extra energy required for AI processing less than the energy it will save? Absolutely yes.”
