Undoubtedly, AI is here, and its adoption is growing at an unstoppable pace. It is booming across nearly every industry. Who today has not used AI in some form? Statista projects that the Artificial Intelligence market will reach a size of US$244 billion in 2025. The market is expected to grow at an annual rate (CAGR 2025-2031) of 26.6%, reaching a volume of US$1.01 Trillion by 2031.

Fei-Fei Li, Professor and AI Researcher at Stanford University, shared at the outset of AI her perspective a few years ago: “As a technologist, I see how AI and the Fourth Industrial Revolution will transform every aspect of people’s lives, impacting how we work, communicate, and solve some of the world’s most pressing challenges.”

AI will disrupt everything. Its predictive, analytical, generative, and interpretative capabilities open up countless applications. This rapid advancement holds both positive and negative implications. Here, we will dive into those related to climate to assess whether this booming technology truly supports the climate transition.

On one hand, the AI boom is coming at an inopportune time from a climate change perspective, shifting the focus of replacing fossil fuel with green energy as soon as possible to suddenly creating a huge additional need for more energy resources for two main reasons:

• Hardware and data centres rely on semiconductors and other technical components that require Rare Earth Elements (REEs), such as gold and tin, along with transformation processes that are often not environmentally friendly, especially for silicium. Additionally, the heat generated, and the intensive cooling requirements of data centres demand large amounts of fresh water, exacerbating their environmental impact.
• Energy consumption of data centres is significant, especially since they require 24/7 reliable power sources. Added to the growing energy demand from electrification and generation constraints, these facilities are adding to the pressure of new supply and some are opting for new fossil fuel generation.

McKinsey predicts that energy demand from U.S. data centres will rise by 10% annually through 2030 and beyond. So US big tech players like Microsoft, Alphabet, and Amazon are investing billions in nuclear as one option to secure energy needs for their growing data centre build out. In October 2024, Google signed a multi-year contract with Kairos Power, a small nuclear reactor (SNR) company shortly followed by Microsoft’s announcement about contracting US$16 billion of power over 20 years to re-open the infamous Three Mile Island nuclear power plant. SNR technologies are unproven and will require years to develop. The race for clean, reliable, 24/7 energy for data centres is on!

Australia offers an ideal location for solar and wind power generation due to its vast land and excellent resources. However, that requires firming for data centres. A growing number of operators are addressing this by integrating on-site solar and battery storage, helping to reduce grid strain and improve resilience. For example, batteries can perform peak shaving, charging when electricity is cheap or solar is abundant and discharging during peak demand, while co-located solar provides clean power with batteries covering cloudy periods. In some cases, such as with Edge Centres in regional Australia, data centres are operating entirely off-grid through solar and batteries. These solutions not only support sustainability goals and reduce reliance on diesel generators but also enhance uptime and unlock new revenue streams through grid services. That, along with new cooling solutions and other related areas are where new technologies can make a big difference.

On the other hand, AI brings significant potential benefits for climate. AI has the power to predict, analyse, compared, create, and innovate thanks to its incredible ability to leverage data.

Many AI-enabled solutions are emerging to tackle climate change, both from a mitigation as well as an adaptation standpoint and both are very much needed.

According to McKinsey energy and material companies had the highest proportion of their digital budget going to generative and analytical AI (17% of them planning to spend 20%+ of their budget).

What is incredible about AI is its versatility, it can be applied to every stage of product development, from hypothesis, testing to observation. Moreover, its outputs can be reused limitlessly, continually refining solutions.

We believe AI has three main promising areas that can support climate action:

• Understanding climate challenges and its effect on earth. Weather forecast is one the greatest application for AI to help avoid dramatic situation or at least help mitigate or prepare for extreme events like recent flood in Spain and fires in Portugal. Sadly, these are no longer rare events but the new norm. For example, Reask, leverages data science and AI modelling to assess and forecast natural catastrophe risks globally, providing critical insights for population but also to support other players like insurance, financial markets, and cities & governments.
• Improving Efficiency with AI Across Key Sectors:
  • Optimising energy systems: To meet climate targets, existing energy infrastructure must operate at peak efficiency. AI is playing a critical role in enabling this transition. It helps detect inefficiencies, optimise grid operations, and balance local energy demand and supply. For instance, companies like GridSight use AI to analyse smart meter data, while Camus Energy leverages real-time grid analytics to orchestrate distributed energy resources more effectively. Despite the rise of AI, McKinsey has noted that energy efficiency has remained flat in recent years, but with advances in model performance and hardware efficiency, this trend may soon shift.
  • Transforming transport & logistics: AI can improve the efficiency of transport networks by optimising routing, reducing idle time, and enhancing load planning. Adiona Tech, for example, uses AI-powered logistics tools to cut emissions and costs for fleet operators by intelligently modelling delivery routes and reducing unnecessary travel. These solutions not only boost operational efficiency but also lower the carbon footprint of last-mile delivery and freight.
  • AI in industrial operations: Industrial systems are often complex and resource intensive. AI tools can help monitor, model, and predict system behaviours to reduce waste and unplanned downtime. Arundo Analytics provides industrial AI software that enables operators to extract insights from equipment and sensor data, improving asset reliability and performance. Such solutions contribute directly to emissions reductions and more sustainable manufacturing practices.

• Developing new materials and biotech: To mitigate climate impacts, we need materials and biological solutions that are environmentally friendly and sustainable. From carbon-neutral materials to bioengineered products that reduce pollution, biotech and advanced material development can reshape industries from agriculture to manufacturing. These innovations have the potential to drastically cut emissions, reduce waste, and build a more sustainable economy. As an example, Matnex is one of many companies using sophisticated AI software to transform the way new materials are discovered and designed

In addition to direct AI solutions, AI is critical in driving startup efficiency. Startups are agile but resource-constrained, so improving operational efficiency is essential for maximising impact. Leveraging AI and data analytics, startups can streamline processes, reduce costs, and enhance productivity. By automating routine tasks and providing strategic insights, startups can focus more on innovation and achieving their climate goals.

A long list of climate-related AI-enable is emerging quickly. Djoann Fal shared the following map which has many interesting companies in the space.

One last thing to consider: scientists, startups, and innovators need quality data to train their models and deploy them effectively in the real world, especially given the high environmental cost of training. To put it in perspective, training a single large language model (LLM) can consume, the equivalent of a year’s energy use in 30 homes, emitting about 25 tons of CO₂ — comparable to a car driving around the world five times, according to Sasha Luccioni’s TED talk. While public data is available and widely used, AI needs access to private and specialised data that addresses specific industry challenges to make a real impact.

At Climate Tech Partners, we believe there are significant benefits in using AI across many areas directly or indirectly helping climate tech startups be much more effective in creating solutions. However, the energy consumption is putting significant pressure on an energy system already overwhelmed by the electrification of the economy and the increased presence of intermittent renewable generation. Many technological solutions will hopefully help mitigate these constraints to enable the full potential of AI with a limited toll on the climate change transition.

References:

• https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai#/
• https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/investing-in-the-rising-data-center-economy