-1-1.png?height=120&name=Curation%20Logo%20Transparent%20(1)-1-1.png)
AI is transforming industries at a breakneck pace, but its energy demands are posing a significant challenge. Global energy startups attracted $9.4bn in funding in 2024, a 12% increase from 2023, reflecting a growing emphasis on sustainable solutions to meet AI’s power needs. Within clean energy options, geothermal saw a remarkable rise in investment, tripling to $558m, while nuclear funding doubled to $1.9bn. As AI becomes increasingly embedded across industries, the need for scalable energy solutions is clear – investment and innovation in climate technology can help facilitate a sustainable future of AI. (Bloomberg)
Why does this matter? Data centres and data transmission networks each account for 1-1.5% of global electricity usage. These centres are crucial for AI workloads but exacerbate energy demands, with models like ChatGPT consuming ten times the electricity of a standard Google search. This surge in energy use can strain grids and undermine ESG objectives, but it also highlights the potential for investors to back transformative solutions in renewable energy, efficient hardware, and cutting-edge infrastructure.
As a leader in the global mining industry, Rio Tinto’s operations generate large volumes of data from autonomous trucks, drills and other advanced systems, all of which are powered by increasingly energy-intensive AI and automation technologies. Addressing the growing energy demands of AI aligns with the company’s commitment to innovation and sustainability.
As AI grows, tackling its energy impact requires innovation across hardware, software and policy. Examples of hardware innovations are emerging, technologies such as Supermicro’s liquid cooling systems are addressing the thermal challenges of high-density AI data centres, improving efficiency while reducing carbon footprints. Similarly, emerging designs such as 3D-stacked memory and in-memory computing have the potential to minimise energy waste by shortening data transfer distances.
In the UK, Octopus Energy invested £200m ($244m) into tech startup Deep Green, pioneering a scheme to recycle the heat from data processing centres into public swimming pools. This has the dual benefit of providing a community service at a reduced cost and taking advantage of the wasted heat from data processing. Such innovations demonstrate AI’s high energy demands can potentially be met sustainably.
A recent analysis by Schneider Electric suggests that software powering AI should be optimised by pruning and quantisation to dramatically cut energy consumption while maintaining performance. Policy-wise, governments and organisations could implement standards for sustainable AI, including energy efficiency certifications and incentives for a circular economy approach. Collaboration between energy providers, policymakers and AI companies is essential to align these strategies.
Renewables are a vital component of decoupling AI growth from environmental degradation. A promising area of rapid renewables innovation is geothermal energy, as seen in Quaise Energy’s partnership with Nevada Gold Mines to retrofit a fossil-fuel plant with geothermal technology, exemplifying the efficacy of co-locating energy generation with business operations. The project could reduce greenhouse gas emissions by 30% while delivering scalable, baseload renewable power. Such innovations demonstrate how renewables can potentially meet AI’s high energy demands sustainably.
Interestingly, AI itself is contributing to solutions for its energy challenges. Machine learning algorithms are being employed to optimise energy usage in data centres, adapting workloads to align with renewable energy peaks and managing cooling systems more efficiently. Beyond AI, these technologies are improving grid resilience by forecasting energy generation and consumption patterns, accelerating the integration of renewables.
AI’s energy demands will only intensify, placing infrastructure innovation and policy responses at the forefront of 2025’s agenda.