· 8 min read
Artificial intelligence is advancing at rapid speed, creating a frontier that requires a careful balancing act: on one hand, AI holds immense potential to reshape our world; on the other, Earth’s life-support systems — climate, biodiversity, and water — are under unprecedented stress and will require ever greater amounts of energy.
One emerging approach in particular, planetary intelligence, offers promise. Planetary intelligence can be defined as a planet’s capacity to sustain habitable conditions through the integration of life, technology, and natural systems. What kind of intelligence are we really investing in? And is there a different way? Will AI learn from nature?
In recent months, ChatGPT surpassed 400 million weekly users. Answering just one prompt now consumes about 10 times more electricity than a Google search. At the same time, OpenAI’s new image generator unleashed a surge of Studio Ghibli-inspired images, fueling the explosion of tools like ChatGPT – which reportedly added 1 million users in just one hour.
The demand for electricity is growing substantially, projected to increase by a robust 3.9% globally in just the next couple of years. It is driven by the electrification of cars, appliances and machinery, and increasing connections to the grid in the Global South as we work towards achieving universal access to energy and 1.18 billion people still live in energy poverty.
But as we pursue energy for human development, AI has turned over the equation, adding a new intensive global demand. Just last year, AI consumed around 415 TWh of electricity, more than 150 individual countries. And although not the same, AI increasingly requires data centers to handle the large electricity load. The demand from data centers worldwide is set to more than double by 2030 to around 945 TWh, which is slightly more than the entire electricity consumption of Japan today.
At the same time, electric grids are being stressed, especially in the Global South. Just a few months ago, Cuba’s power grid collapsed, triggering a nationwide outage and plunging more than 10 million people into darkness. Much of Havana now faces near-daily power cuts lasting four to five hours, and in rural parts of Cuba, sometimes 20 hours or more.
Additionally, if you speak to local governments, the water usage from data centers presents a serious planning concern. AI and the data centers it so often requires are projected to consume up to 6.6 billion cubic meters by 2027, enough to grow food for more than 6 million people each year.
Last year, in Querétaro, Mexico, one of the regions hardest hit by drought, farmers rationed water and lost crops. Just a few kilometers away, one AI data center was given permission to extract 25 million liters of groundwater per year for just a single unit. That is 24% of all the water allocated to the entire municipality for public use. At the same time, around 70% of the world’s lithium, cobalt, copper, and rare earth elements, the minerals powering AI and clean energy infrastructure, come from the Global South, yet much of the value is extracted elsewhere. Global e-waste hit 62 million tons in 2023, with less than 25% recycled and much of it dumped back into those same regions.
------------
What makes AI such a paradox to energy systems is that while it presents serious resource concerns, it can also increase efficiency in energy systems, helping to reduce waste and prepare for growth in electricity demand, partially fueled by AI.
For example, generative AI can improve forecasts for electricity supply and demand, helping to increase efficiency. It can predict faults in energy systems ahead of time, making grids more stable. It can also enable other efficiency technologies, such as smart meters, which send real-time energy consumption patterns to utilities. In Indian states such as Uttar Pradesh and Assam, smart meters have already achieved technical loss reductions of up to 20 percent. Through these and other efficiency gains, AI could reduce global greenhouse gas emissions by up to 10% by 2030. That is roughly equivalent to the annual emissions of the entire European Union.
Therefore, when generative AI is integrated with smart meters, it can also support the expansion of decentralized energy systems such as rooftop solar. This creates new economic opportunities and democratizes energy systems, which was difficult with fossil fuels. Oil, gas and coal supply require far more infrastructure and bureaucracy.
It is, of course, also essential that AI developments are powered by sustainable energy sources.
Without access to energy, health centers cannot power refrigeration for vaccines, the elderly and sick die from heatwaves without air conditioning, children do not have lights to study at night, and women may spend hours of their day collecting firewood as a fuel source.
While advances in AI will eventually make it more energy and water efficient, the extent and timeline remain subjective. Given this, and energy’s catalytic role in enabling development, it is critical that we ask right now:
Are we using AI in a way that advances sustainable energy for all? Or are we just adding another hefty resource demand for the most privileged energy consumers, who already have their basic needs met?
---------
The answer can come from nature itself. A re-thinking on how AI is powered, governed, and integrated into the systems that sustain life, stopping seeing nature as something to fix, and started seeing it as something to learn from.
For example, biomimicry, or taking product design inspiration from nature, has gifted us solar panels that self-clean like lotus leaves; wind turbines modeled after whale fins; and buildings that mimic termite mounds to regulate indoor temperature.
But the deeper opportunity lies not just in copying nature’s design, but internalizing nature’s relationships. That’s the promise of planetary intelligence, a new development paradigm that seeks to train AI based on nature’s patterns of interdependence. Biomimicry innovations could unlock $1.6 trillion in economic value by 2030, by boosting efficiency in sectors like water, energy, electronics, and mobility. But what if we elevated biomimicry from product design to planetary design?
Nature doesn’t just offer us clever components. It offers us whole systems—ecosystems that are regenerative, interconnected, and adaptive. AI can begin to learn from these deeper logics. Too often, energy, nature, and AI have been approached in silos. However, as generative AI increasingly functions as the operational brain for electric grids – forecasting, optimizing, and balancing – it can play a more holistic role in managing tradeoffs across water, energy, land, and biodiversity, and could generate up to $1.3 trillion in economic value by 2030.
For example, electric grids and renewable energy sources do not exist in siloes. Rather, they can be found near farms and forests; neighborhoods and rivers. AI could manage biodiversity concerns from wind farms by tracking flight patterns of birds in the area. Similarly, during heatwaves, which are becoming more frequent with climate change, AI can help cities manage grid strain, coordinate water use, and protect urban green spaces.
In this way, AI can help us see the forest and the trees. If AI is trained based on the principles that govern natural systems, it could support the creation of more sustainable and resilient ecosystems.
------------
Yet while planetary intelligence and AI overall present opportunities for sustainable energy development, its distribution remains deeply unequal. Of the projected $15.7 trillion in global economic value from AI by 2030, approximately 84% is expected to accrue to just three regions: China, the United States, and Europe.
Only 1% of the world’s most powerful supercomputers are located in Africa. Many countries in the Global South face limitations in digital infrastructure, data sovereignty, and institutional capacity, constraining their ability to shape, deploy, or benefit from advanced AI systems. Additionally, the vast majority of large generative AI models continue to be trained on datasets that overlook non-Western languages, cultures, and contexts. An estimated 2.6 billion people still lack internet access and remain effectively invisible to the AI systems that increasingly shape decision-making in sectors such as health, finance, and public administration. Initiatives such as the Masakhane project are addressing this gap by developing natural language processing models in over 40 African languages, offering a template for more representative and culturally grounded AI development. Thus, it is essential to frame a just transition around energy and AI around two things: first, that AI works to further sustainable energy and nature goals; and second, that all countries can access the AI that enables this.
The governance of AI must be approached with the same foresight and care applied to other critical systems such as water, forests, and biodiversity. Without robust ethical frameworks, AI is already being weaponized for harmful purposes — ranging from optimizing illegal logging operations to enhancing fossil fuel extraction and surveilling environmental defenders.
Nevertheless, the global landscape is shifting. Countries in the Global South are emerging as leaders, developing frameworks for AI’s regulatory use and collaborating with each other for context-specific technologies. It is not a question of if, but how and when.
illuminem Voices is a democratic space presenting the thoughts and opinions of leading Sustainability & Energy writers, their opinions do not necessarily represent those of illuminem.
Track the real‑world impact behind the sustainability headlines. illuminem's Data Hub™ offers transparent performance data and climate targets of companies driving the transition.
