AI’s Power Surge: Transforming Innovation In The Energy Sector – IEA

Artificial intelligence (AI), like the steam engine and electricity, is a general-purpose technology that has the potential to transform the global economy and the energy sector. While uncertainties remain, AI could play a significant role in accelerating innovation in energy technologies. Recent advancements have reduced the costs of key technologies, but achieving global energy security and emission targets requires further improvements and the development of new solutions. AI can enhance scientists’ ability to generate and test new ideas, but for this potential to be fully realized, collaboration between policymakers and the scientific community is essential.

The International Energy Agency (IEA) is focusing on AI’s role in energy innovation through a new workstream that examines its impact on areas such as electricity consumption in data centers and optimizing complex systems like power networks. The upcoming Global Conference on Energy & AI will bring together leaders from government, the energy sector, and the tech industry to discuss how AI can drive progress in these areas.

A fundamental question for energy analysts is whether AI can accelerate the pace of technological advancements. In many industries, progress has followed predictable patterns, such as Moore’s Law in semiconductors, which predicted the doubling of transistors every two years. For energy technologies, cost reductions are often linked to cumulative deployment, known as the “learning rate.” For example, the learning rate for electric vehicle batteries is around 15%. However, factors like supply chain constraints and rising costs of materials have raised concerns about maintaining these rates in the future.

Some experts believe AI can help sustain or even accelerate these learning rates, while others view it as a disruptive force that could surpass current projections. AI’s potential lies in its ability to analyze vast datasets and optimize complex problems. Early examples of AI-driven discoveries in energy materials are promising. In July 2024, researchers from a U.S. government lab and Microsoft used AI to evaluate 32.5 million potential solid-state electrolytes for lithium batteries, identifying 23 with desirable properties. Similarly, scientists in Sweden screened 45 million potential battery cathode molecules, finding nearly 4,600 promising candidates. Startups like Anionics and Mitra Chem are leveraging AI to speed up the development of new materials, attracting significant investment.

AI’s contributions are not limited to batteries. It has been used to engineer enzymes for biofuels, predict high-yield biofuel feedstocks, and identify catalysts for hydrogen production and carbon capture. AI tools have also enabled researchers to create materials for capturing carbon dioxide. As AI becomes integral to energy research, advancements in robotics and automation will further boost innovation. A recent study showed that using AI in an industrial setting led to a 39% increase in patenting within two years.

Despite these successes, challenges remain. One major issue is data availability. AI models require large, structured datasets, but existing databases often lack comprehensive information. Efforts are underway to expand datasets like the Materials Project and Cambridge Structural Database. While synthetic data can fill some gaps, real-world experimental data is crucial. International collaboration, such as the Mission Innovation M4E platform, could help develop standardized protocols and shared datasets.

Another challenge is optimizing materials for multiple characteristics and ensuring they are suitable for practical use. AI can design materials, but human expertise is still needed to test performance under various conditions and develop manufacturing processes. Expanding AI’s role in these areas will require significant computational resources and investment.

Even if AI accelerates discovery, the process of prototyping, commercializing, and scaling up new technologies remains time-consuming. Innovations like self-driving labs could help. For example, the A-Lab at Lawrence Berkeley National Laboratory uses robots to synthesize chemicals predicted by AI, processing up to 100 times more samples than human-run labs. Digital twins, virtual models of facilities or processes, are also being used to optimize complex systems, such as nuclear fusion and CO2 capture.

However, access to these advanced tools is limited, and skills gaps may hinder progress. Policymakers need to support the development of digital tools, invest in training, and create regulatory frameworks that facilitate innovation. Ensuring that AI-driven solutions reach the market will require cooperation between governments, innovators, and investors.

AI has the potential to revolutionize energy innovation, but its success depends on addressing key challenges. Investments in data, skills, and technology, along with international collaboration, will be critical. If these efforts are successful, AI can accelerate innovation, reduce costs, and enhance economic competitiveness. By improving decision-making and optimizing new technologies, AI can create value for consumers and drive progress in the energy sector.