The role of AI in the energy transition

TL;DR

• The energy transition, and just energy transition, are complex processes with various factors feeding into them. Opinions on the exact composition of these terms vary.  

• Artificial intelligence is impacting several of these factors, serving to drive energy transition progress forwards.  

• From a technical perspective, AI is enabling the development of smart grids and predictive maintenance – both crucial to successfully onboarding renewable power to energy systems.  

• Economically, AI is assisting with commercial market analysis and risk management, helping to make the financial case for investing in the energy transition more robust.    

• Socially, AI is enabling consumers to alter their energy consumption behaviour by providing real-time insights into usage patterns.  

• AI is also enabling transparency and accountability in relation to energy developments in local communities, as well as providing career transition opportunities for the energy workforce.  

• However, amid all of the benefits comes a cost that few are aware of. AI, and the data fuelling it, consumes enormous amounts of electricity and water every year. 

The detail

What do we mean by energy transition?  

The term is bandied around a lot and means different things to different people. Interpretations may vary depending on where you are in the world, what industry you work in, or even how you go about living your everyday life. 

Personally, I see it as a process or journey made up of multiple strands. Three are of particular importance.  

First is the technological shift. Here, energy transition refers to the adoption of alternative energy sources such as solar, wind, hydroelectric power and nuclear to replace traditional fossil fuels. Alongside this, technologies designed to boost energy efficiency are being developed and deployed to reduce the burden on power grids.  

Second is an economic transformation. By this, the energy transition signifies a restructuring of industries and economies towards low-carbon energy production, fostering innovation, job creation and competitiveness in the global market.  

Third is social change. To be successful, energy transition requires societal shifts in behaviour and consumption patterns, promoting energy efficiency, conservation and equitable access to clean energy resources, while fostering a more sustainable and inclusive future. 

This leads to another concept frequently discussed: the just energy transition. Advocates of a just energy transition stress the importance of fairness and social justice throughout the shift to sustainable energy. They aim to ensure that vulnerable communities reap the benefits of the transition and aren't unfairly burdened by its costs or adverse effects. 

In particular, a just transition should focus on engaging local communities in decision-making processes related to energy projects, empowering them to have greater control over energy resources and infrastructure. From a labour standpoint, it should also prioritise safeguarding workers' rights and livelihoods during the transition to clean energy. This includes supporting workers in fossil fuel industries through retraining, job creation, and fair labour practices. 

To fulfil all these ambitions, multiple stakeholders must come together at local, national and international levels. Governments, communities and the private sector each have an enormous role to play.  

But it is not just human traits such as willpower, ingenuity and selflessness which will win the day. Artificial intelligence (AI), and how we use it, will also have a major impact on the speed and success of any energy transition journey. 

How is AI impacting the energy transition?  

AI touches on many of the factors I have just described as feeding into the meaning of energy transition.  

From a technological perspective, AI is already playing a crucial role in incorporating renewable power into energy systems and grid infrastructure.   

As the demand for electricity grows and efforts to reduce carbon emissions escalate, power systems are undergoing a significant transformation. Once reliant on centralised power stations, modern grids must now navigate a complex web of energy flows between distributed generators, the grid and end-users – from individual homes to industrial facilities, and many other end points in between. Meanwhile, the proliferation of grid-connected devices, be it EV charging stations, rooftop solar panels or otherwise, adds unpredictability to these flows.  

All of this shines a spotlight on the need for enhanced information exchange and powerful tools for system planning and operation. This is where AI is proving its worth. With machine learning models doubling in computational power every few months since 2010, AI can now perform tasks like language recognition, image analysis, and even self-programming.  

Given these capabilities, it is no surprise to see the energy sector is tapping into AI's potential to drive efficiency and innovation. AI is uniquely equipped to handle the data deluge from smart grids, where smart meters generate thousands of times more data than traditional devices. Massive volumes of data are being created at the other end of the spectrum, too. For instance, the world’s wind turbines have been estimated to produce more than 400 billion data points per year.  

AI is the only form of intelligence capable of handling this level of information. According to analysis carried out by Indigo Advisory, the market for AI in the energy sector could already be worth up to $13 billion, this value being derived across more than 50 possible use cases.  

Another one of those use cases is demand forecasting. AI-based forecasting models can accurately predict renewable energy production, such as solar irradiance and wind speeds, enabling better integration into the grid and planning of energy generation. This is already being applied in the real world, with Google’s software being piloted by Engie.  

At the same time, AI can facilitate demand response systems, adjusting energy consumption patterns in response to real-time pricing or supply fluctuations, thus reducing peak loads and enhancing grid stability.  

Predictive maintenance techniques will also be a critical part of the energy transition journey. By continuously monitoring the performance of key assets, AI-driven solutions can reduce downtime and increase the lifespan of energy infrastructure, such as wind turbines and solar panels, ensuring optimal performance.  

The same applies to grids, where utility companies around the world have been adopting preventative maintenance systems.  

E.ON is a good example. The company has created a machine learning algorithm that forecasts when medium voltage cables within the grid require replacement. This algorithm analyses data from various sources, pinpointing patterns in electricity generation and detecting any anomalies. According to E.ON's findings, employing predictive maintenance techniques could potentially slash grid outages by up to 30% compared to traditional methods. 

AI and the economic and societal shift 

The technical impact of AI on the energy transition is already huge and going to get bigger. I could go on listing examples, but it’s important not to forget the other components of the energy transition I mentioned at the beginning, as AI is playing a key role here as well.  

The economic transformation is being powered by AI in several ways.  

Market analysis is one use case. Here, AI can analyse key energy sector trends, enabling better decision-making for investments in renewable energy projects and facilitating the transition to a low-carbon economy. In a similar vein, AI can also aid with risk management processes by assessing and mitigating risks associated with energy investments – these include regulatory changes, market fluctuations and climate impacts.  

We also talked about the just energy transition and the emphasis on job creation and protecting livelihoods. AI is again having a significant impact here. Its deployment across the energy ecosystem is creating new job opportunities in technology development, data analysis and system optimisation, contributing to employment growth and skills development in the workforce. 

Another critical aspect of a just energy transition is equity in access. To support this, AI algorithms can analyse demographic and geographic data to identify areas with limited access to clean energy resources. This information can be used to prioritise investment in renewable energy infrastructure and energy efficiency programmes in underserved communities. 

Transparency and accountability are also important. AI-driven monitoring systems can collect and analyse data on energy projects, including their environmental and social impacts. For example, satellite imagery and sensor data can be leveraged to monitor pollution levels in different neighbourhoods, identifying areas with disproportionate environmental burdens.  

I also highlighted social change and shifts in behaviour as being key to a successful energy transition journey. Here, AI can analyse consumer behaviour data to develop targeted energy efficiency campaigns, encouraging individuals to adopt more sustainable energy practices and reduce consumption. 

Is there an elephant in the room? 

The case for AI to drive the energy transition forwards is undeniable. There are boundless use cases and applications – too many to go into detail in just one newsletter. 

The rapid pace of progress and widespread adoption of AI, extending far beyond energy and encompassing nearly every facet of life and business, creates a sense of momentum that seems nearly unstoppable. However, little attention is paid to the enormous environmental footprint that AI itself creates.  

AI feeds on data, and to say there is a lot of data being produced around the world is an understatement. We talk about the ‘datasphere’ in zettabytes now – that’s a trillion gigabytes at a time. In 2010, there were around 0.5. zettabytes of data in existing. By 2025, it is estimated that there will be over 50.  

All this data needs to be stored somewhere. According to the Internation Energy Agency (IEA), data centres account for up to 1.5% of global electricity use. Data transmission networks also account 1-1.5%. In terms of carbon, the two combine to be responsible for 1% of energy-related emissions.  

Water is another resource data centres rely on to function. In 2021, Google’s data centres consumed around 4.3 billion gallons of water to keep its servers cool. According to ING, a mid-sized data centre in the US will use around 300,000 gallons of water a day, which is the same as around 100,000 homes.  

While I am not advocating the winding down of data centres and switching off AI (you simply can’t), it is important to consider the physical footprint associated with its exponential growth. Energy efficiency improvements are being made, and 1% of emissions may appear a small price to pay given the progress that AI is contributing to. As the energy transition journey continues, so too must the conversation around the resources being used to power those technologies working away in the background.  

— Lew 👋

As ever your feedback is important to me. Please help by letting me know what you love or what you think can improve.

The Transition’s work is provided for informational purposes only and should not be construed as advice in any capacity. Always do your own research.