Watt’s up with AI in cleantech?

Explore how AI is making strides in cleantech, enhancing efficiency, cutting costs, and driving innovation.

Bottom line

With vast amount of data and need for real-time interventions, AI has extensive applications in the cleantech sector, spanning renewable energy generation, power grid optimization, material science, and climate change monitoring. Furthermore, the increased use of AI in cleantech is expected to address many challenges the industry is facing.

What happened

Recently, there has been a lot of talk about how AI (and datacenters) are driving up electricity demand, with investors paying close attention to the potential positive impacts AI can have on utilities, renewables deployment, and even nuclear and gas demand.

However, AI is not just consuming energy; it is also a transformative tool poised to drive efficiency improvements across the cleantech sector.

Many cleantech applications are prime candidates for AI due to two key characteristics: the vast amount of data available and the need for real-time adjustments. For example, it becomes possible to optimize the use of renewable energy sources (like solar and wind) by continuously collecting data on weather patterns and grid conditions. Processing and acting on this data swiftly allows for real-time adjustments where human intervention might fall short.

While AI is not new to cleantech, its application is increasingly relevant given the sector's specific needs and challenges. It is being used to make renewable energy systems smarter and more efficient, optimizing everything from the performance of solar panels and wind turbines to the management of energy grids. This means higher productivity and lower maintenance costs for renewable energy sources.

Moreover, AI could play a key role in monitoring and combating climate change. By analyzing massive datasets from various sources, AI could help us track environmental changes, predict climate-related disasters, and develop effective adaptation strategies.

In the next section, we will delve deeper into how these developments may impact our investment strategy and the broader cleantech landscape.

Impact on our Investment Case

Forecasting the future

As renewable power generation continues to grow, it becomes increasingly important to accurately predict the output from solar and wind resources and forecast power demand. Today, variable renewables (solar and wind) represent roughly 12% of global electricity generation, a share expected to grow to 25% by 2028, according to the IEA’s latest report. With more variable renewables in the electricity generation mix, predicting how much power will be produced at any given time becomes crucial.

This is where AI comes into play. Machine learning (ML) and deep learning (DL) algorithms trained on historical weather data and climate patterns significantly improve the precision of energy output predictions for solar and wind installations (by up to 25% compared to traditional methods). This enables operators to better anticipate production fluctuations and maintain a stable power supply. 

New ML/DL algorithms, which use multiple weather inputs such as wind speed, temperature, humidity, rainfall, and solar irradiance, are being developed. Some studies show that such methods enable better prediction for solar radiation and wind speeds compared to traditional methods, leading to better energy management and enhanced grid stability.

AI also plays a crucial role in optimizing the siting and design of renewable energy projects. By analyzing geographic and meteorological data, AI helps identify the best locations for solar and wind farms, maximizing energy generation efficiency. This includes assessing factors like sunlight intensity, wind patterns, and proximity to existing grid infrastructure.

SAS Industries, for instance, has developed a platform for AI-powered energy forecasting, enabling utilities to operate more efficiently, optimize power trading, and better select locations for new generation  capacity.

On the power demand side as well, AI has the potential to help improve predictions. Indeed, AI models can help forecast short-term and long-term energy demand by analyzing consumption patterns and other influencing factors (e.g., economic activities, population growth, etc.). One Swiss startup active in this field is Hive Power, whose solutions use AI to monitor resource distribution and analyze consumer usage behavior. This capability reduces the risk of blackouts and ensures that renewable energy is utilized to its full potential.

Smart grids get smarter

Reaching net zero emissions requires not just more renewable energy but a smarter, more efficient grid. Indeed, it is estimated that annual grid investment must double from the current $300bn globally to $600bn by 2030. Beyond purely infrastructural investments in new transmission and distribution lines, we believe AI will play a crucial role in enhancing grid operation and management.

Picture the grid as a complex nervous system that requires constant fine-tuning and adaptation. Intelligent systems act as the brain, processing real-time data to optimize energy flow and maintenance, making our existing infrastructure more efficient and delaying costly new investments.

One way AI can help is through dynamic line rating (DLR), an advanced technique for optimizing the capacity of transmission lines by adjusting their power-carrying limits (called "ratings") based on real-time weather conditions. Traditionally, transmission lines were rated using conservative assumption (worst weather scenario), leading to underutilization as the lines are often capable of carrying more power under normal conditions.

Incorporating AI and DLR allows grid operators and utilities to use real-time weather data (e.g. temperature, wind speed, solar radiation, etc.) to dynamically adjust these power limits and optimize grid capacity, something that would be impossible to implement with human-based controls. This approach enhances grid efficiency, reduces energy wastage, and allow lines to carry more electricity safely. Some studies have shown that DLR can increase the capacity of transmission lines by 20% to 30%, depending on conditions.​  An MIT study published in 2022 found that incorporating DLR on the ERCOT Transmission System could save $776 million annually mostly through reducing their congestion rents. Congestion rents occur when electricity demand exceeds grid capacity, leading to higher transmission prices as ERCOT has to use alternative routes or additional resources to manage the load. By improving grid efficiency, these costs are significantly reduced. In 2023, ERCOT's total congestion rents were $2.8 billion. 

AI is also key in enhancing the integration of distributed energy resources (DERs) such as rooftop solar panels and small wind turbines. These resources generate power at the edges of the grid, making it more complex to manage. Intelligent coordination ensures that these decentralized sources work seamlessly with the larger grid to maintain stability and efficiency. Itron Inc is making some progress in this area with its Distributed Energy Resource Management System (DERMS) that enhances grid resilience by pinpointing DER locations, sensing local grid conditions, and offering control.

Additionally, Stem provides AI-driven clean energy solutions through their Athena platform. Athena optimizes energy storage and integrates distributed energy resources into the grid. By using real-time data analytics, Athena helps to balance supply and demand, improving grid stability and efficiency.

Financially, these innovations in grid management reduce operational costs and delay/ reduce the need for expensive infrastructure projects.

Keeping the power flowing

Maintaining the reliability of renewable energy infrastructure is crucial for harnessing its full potential. Predictive maintenance, powered by AI, is revolutionizing this field. By proactively identifying potential equipment failures before they occur, predictive maintenance helps reduce unplanned downtime and extend the lifespan of renewable energy assets.

In solar energy, AI-driven predictive maintenance monitors photovoltaic (PV) panels in real-time. By analyzing data from embedded sensors, such as temperature and performance metrics, AI algorithms detect anomalies like reduced efficiency or degradation. Early detection allows operators to schedule maintenance before significant performance issues arise, ensuring optimal energy production and prolonging the lifespan of installations. Swiss startup SmartHelio, for example, uses AI to analyze real-time data from solar panels, identifying issues early and optimizing maintenance schedules. According to their claims, this approach could provide an overall improvement in solar asset performance by 15% and a reduction in operation and maintenance (O&M) costs by more than 50%.

Wind turbines, which face varying wind speeds and environmental conditions, are perfect candidate to AI-powered maintenance strategies. Components like bearings and gears are prone to wear and tear. AI analyzes sensor data on vibration, temperature, and other indicators to predict when these components will likely fail. By scheduling maintenance proactively, operators can minimize downtime and maximize energy production, improving the overall efficiency of wind farms.

Additionally, digital twins (virtual models of physical assets) play an increasingly important role in predictive maintenance. These digital replicas allow operators to simulate and analyze the condition of their equipment in a virtual environment, leading to more accurate predictions and efficient maintenance planning. Both Siemens Gamesa (now part of Siemens Energy AG) and GE Vernova use digital twin technology to monitor and predict turbine performance, allowing for proactive maintenance and optimization of energy output (reducing up to 20% in maintenance costs according to their documentation).

Enphase and SolarEdge are also making strides in predictive maintenance. Enphase’s Enlighten monitoring platform provides comprehensive analysis of solar production, enabling remote maintenance and management of solar assets. SolarEdge’s predictive maintenance capabilities, integrated into their Energy Hub inverter, help identify potential issues before they cause significant disruptions.

While implementing predictive maintenance requires some upfront costs (for sensors, data infrastructure, etc.) the long-term benefits of reduced downtime, lower maintenance costs, and extended equipment lifespan make it a worthwhile investment.

Material breakthroughs

Materials innovation is crucial for decarbonization, and AI is significantly accelerating this process which traditionally relied on trial and error. AI models predict properties of materials based on their chemical structures, screening millions of possibilities to find the best candidates for energy applications. This significantly speeds up the discovery process and reduces costs.

A standout example is the recent breakthrough at EPFL, where machine learning was used to discover new perovskite materials for solar cells (perovskites are a class of materials that can be used to create efficient and potentially low-cost solar cells). These lead-free perovskites promise higher efficiency and durability, featuring properties like direct band gaps (which allow for better light absorption) and low exciton binding energies (which improve charge separation and transport). The ability to identify suitable materials quickly and accurately could typically enhance local manufacturing capabilities by reducing reliance on specialized supply chains and complex production methods.

At the same time, AI is driving innovation in solid-state battery technology, which many view as the future of energy storage due to its higher energy density and improved safety. Microsoft Corp, in collaboration with the Pacific Northwest National Laboratory (PNNL), recently leveraged generative AI to screen over 32 million potential materials for optimal solid-state battery electrolytes. This exhaustive search, which would have taken years without AI, was completed in less than 80 hours, ultimately identifying 23 promising candidates, only five of which were previously known.

The researchers used Microsoft’s Azure Quantum Elements (AQE) platform, which combines high-performance computing (HPC) and AI, to query for battery materials that use less lithium. Initially, the AI suggested 32 million candidates. It then discerned around 500,000 as stable enough for further consideration. Additional filtering narrowed this down to the final 23 candidates. One of these candidates was synthesized and tested, successfully powering a lightbulb and clock. While this is a promising start, much more work is needed before these materials can be used in electric vehicles. Notably, this specific material could reduce the lithium required in batteries by up to 70%.

Material innovation is crucial for the energy transition, and AI has the potential to discover new material structures, potentially reducing our reliance on a few countries that dominate the extraction and processing of key resources.

Climate guardians

Beyond improving efficiency in clean technologies, AI is also being leveraged to tackle climate change directly through advanced climate modeling, disaster prediction, and greenhouse gas (GHG) emissions monitoring.

AI enhances climate modeling by processing vast datasets to make accurate predictions. The European Space Agency’s "Destination Earth" initiative uses AI to create a digital twin of Earth, helping predict climate phenomena and inform adaptation strategies​.

For disaster prediction, AI analyzes historical and real-time data to forecast hurricanes, floods, and wildfires. This enables timely interventions and better disaster management. AI models are used to map wildfire risks and track their development in real-time, aiding efficient disaster readiness​.

In monitoring GHG emissions, AI combined with advanced satellite systems offers a robust solution for real-time data analysis and environmental management. Climate TRACE, for instance, uses satellite data to track emissions from various sources, enhancing regulatory compliance and pinpointing areas for emission reductions. Companies such as privately held SpaceX and publicly listed Rocket Lab USA Inc could greatly benefit from this, as the demand for sophisticated satellite technology to support these AI-driven monitoring systems grows. 

Sentinel-2 satellites (2A, 2B, and the upcoming 2C) from the European Space Agency's Copernicus program provide high-resolution imagery using their 12 spectral bands. These satellites can detect and analyze methane emissions and other pollutants accurately.

Additionally, NASA's Orbiting Carbon Observatory-2 (OCO-2), operational since 2014, measures atmospheric CO2 levels with high precision. OCO-2 captures millions of measurements daily, offering detailed insights into CO2 sources and sinks, which is essential for developing effective climate strategies.

Integrating AI into climate modeling, disaster prediction, and emissions monitoring helps make informed decisions that enhance climate resilience.

Weighing the pros and cons

While AI significantly enhances clean technology and climate initiatives, it also introduces several risks that need careful consideration to ensure a balanced approach.

Greenhouse Gas Emissions

While AI can optimize processes to reduce emissions, its own carbon footprint is a concern. The energy-intensive nature of training and running AI models contributes to greenhouse gas emissions, especially if the energy comes from non-renewable sources. Innovations in energy-efficient hardware and ensuring the use of renewable energy sources are essential to mitigate this impact.

Security Risks

AI systems bring new security vulnerabilities by expanding the "attack surface". They are susceptible to adversarial attacks, where malicious actors manipulate AI models by tampering with the data they rely on, leading to incorrect outputs. Ensuring robust security measures is crucial to prevent these risks, especially in critical infrastructure like power grids, where failures could have severe consequences.

Considering this, we anticipate a significant rise in the use of the Internet of Things (IoT), AI, and enhanced cybersecurity measures. IoT devices, such as sensors and smart meters, collect real-time data on energy consumption, production, and distribution, which is crucial for understanding patterns and anomalies within the grid. AI algorithms analyze this vast data to predict energy demand, detect faults, and optimize energy distribution. However, the interconnectedness of IoT devices makes them vulnerable to cyber-attacks. AI can monitor network traffic in real-time to detect and respond to potential security threats, safeguarding the grid’s infrastructure. Companies like CrowdStrike benefit from the increased use of AI in the grid by providing advanced cybersecurity solutions.

Ensuring that such critical infrastructures have appropriate security in place is crucial, a key area of interest for our Cybersecurity strategy.

Privacy Concerns

The extensive data collection necessary for AI applications raises significant privacy issues. Sensors and IoT devices used to monitor environmental and human activities can lead to unauthorized surveillance if not managed correctly. The population is quite sensitive to such topics. For example, the Linky controversy in France highlighted concerns over data leakage from smart meters, potentially revealing detailed personal habits. This underscores the importance of implementing stringent privacy safeguards to prevent misuse and protect individual privacy. 

Our Takeaway

Yes, AI consumes significant energy, with datacenter energy usage projected to increase from 2% of global electricity consumption in 2022 to roughly 4% by 2026. However, this doesn't mean AI is detrimental to the environment. A substantial portion of this power can come from renewable sources (as we highlighted in a previous note). Furthermore, as detailed in this article, AI has numerous applications that foster the energy transition, such as integrating more renewables, making smart grids more efficient, and extending the lifetimes of generation assets through predictive maintenance.

While it is difficult to predict precisely how much energy AI will consume in the future, it is equally challenging to quantify the efficiency gains AI-driven technologies will bring. We cannot exclude the possibility of rebound effects, where efficiency gains lead to higher overall consumption (known as the Jevons paradox). However, it is clear that AI must and will be adopted to enhance clean technologies positively. According to a World Economic Forum report published in 2021, a 1% demand-side efficiency gain achieved by AI could save $1.3 trillion in clean energy investments from 2020 to 2050. Additionally, AI can help grid operators optimize power transformer usage, potentially saving $188 billion in costs (over that same period) associated with increased air temperatures impacting transformer lifespans. The report also highlights that a lack of intelligent flexibility could increase power system costs by 6-13% in 2040.

Not all AI applications have a positive impact on the environment, but those described in this article will likely be deployed on a much larger scale in the coming years to accelerate the energy transition and reduce associated costs. Our Sustainable Future strategy is well-positioned to capitalize on this trend.



This report has been produced by the organizational unit responsible for investment research (Research unit) of atonra Partners and sent to you by the company sales representatives.

As an internationally active company, atonra Partners SA may be subject to a number of provisions in drawing up and distributing its investment research documents. These regulations include the Directives on the Independence of Financial Research issued by the Swiss Bankers Association. Although atonra Partners SA believes that the information provided in this document is based on reliable sources, it cannot assume responsibility for the quality, correctness, timeliness or completeness of the information contained in this report.

The information contained in these publications is exclusively intended for a client base consisting of professionals or qualified investors. It is sent to you by way of information and cannot be divulged to a third party without the prior consent of atonra Partners. While all reasonable effort has been made to ensure that the information contained is not untrue or misleading at the time of publication, no representation is made as to its accuracy or completeness and it should not be relied upon as such.

Past performance is not indicative or a guarantee of future results. Investment losses may occur, and investors could lose some or all of their investment. Any indices cited herein are provided only as examples of general market performance and no index is directly comparable to the past or future performance of the Certificate.

It should not be assumed that the Certificate will invest in any specific securities that comprise any index, nor should it be understood to mean that there is a correlation between the Certificate’s returns and any index returns.

Any material provided to you is intended only for discussion purposes and is not intended as an offer or solicitation with respect to the purchase or sale of any security and should not be relied upon by you in evaluating the merits of investing inany securities.