As artificial intelligence (AI) continues to shape the development of society, critical questions are being raised about the technology’s sustainability. Does AI have the potential to solve our most pressing climate challenges—or are we at risk of it exacerbating environmental problems?
AI and Resource Consumption
Artificial intelligence is not just advanced algorithms—it’s a technology that demands massive resources. The data centers powering AI models consume:
- Electricity
Enormous amounts of energy are required to run and cool servers, much of which still comes from fossil fuels. - Water
Used for cooling the servers. According to researcher Shaolei Ren, up to half a liter of water evaporates just to answer a set of 4–5 questions, highlighting the resource intensity of AI systems. - Minerals
Rare metals like cobalt and lithium, along with rare earth elements like neodymium and dysprosium, are used in the hardware. Extracting these resources can lead to severe environmental damage and human rights violations.
A 2019 study (Strubell et al.) showed that training a single AI model could emit as much CO₂ as a car does over its entire lifetime. This demonstrates how energy-intensive AI can be—especially when powered by non-renewable sources.
The Minerals Behind the Machines
Another sustainability challenge lies in the growing demand for raw materials. Sophisticated AI systems require increasingly advanced hardware, placing pressure on resources such as:
- Rare earth elements like neodymium and dysprosium, commonly used in magnets for electric motors and turbines, as well as in high-performance computers.
- Critical metals like cobalt and lithium, essential to battery technology, have become highly sought-after due to electrification and digitalization.
Mining for these resources can have devastating effects on ecosystems and, in countries with weak regulations, may result in human rights abuses. So where is the sustainability in all of this?
Can AI Promote Sustainability?
Despite its challenges, AI also holds vast potential to support sustainability efforts:
- Energy optimization:
AI can predict energy consumption patterns and help use renewable energy more efficiently. Smart energy systems can balance supply and demand, reducing waste. - Waste reduction:
In the food industry, AI can analyze production, inventory, and demand to reduce food waste. Similarly, industrial processes can be optimized to minimize material waste. - Environmental monitoring:
Using satellite data and advanced analytics, AI can track deforestation, map ocean pollution, and follow the migration of endangered species in real time. This supports governments and organizations in making informed decisions for conservation and regulation. - Sustainable urban planning:
Cities account for a large portion of global greenhouse gas emissions. AI-based systems can help plan efficient public transport, coordinate traffic, and develop smart buildings that consume less energy.
An Industry Out of Balance
Even with all its potential, the AI industry faces major challenges when it comes to sustainability and resource use. Microsoft’s water consumption rose by 34% in 2022—enough to fill 2,500 Olympic-sized swimming pools—partly due to its partnership with OpenAI.
A more recent example pointing in a different direction is the Chinese company DeepSeek, founded in 2023. On January 20, 2025, it launched its AI model R1, which drew attention for its low cost and resource use compared to competitors. Despite rivaling OpenAI’s models in performance, R1 runs on simpler hardware and consumes less energy—indicating a possible industry shift toward more cost-efficient and potentially sustainable solutions.
DeepSeek has also made its model open source, which may foster innovation and transparency in an industry often accused of greenwashing. At the same time, the company’s approach highlights how politics shape technology. For example, the model avoids sensitive topics such as Taiwan’s political status, instead focusing on technical issues. This illustrates how the AI industry must constantly balance technology, politics, and resource consumption.
As SINTEF researcher Signe Riemer-Sørensen notes, lack of transparency is a widespread issue in the AI field. When tech companies promise sustainability, we must ask: is it genuine innovation—or an attempt to hide the environmental costs?
The Path to a Greener AI Future
For AI to become a positive force in the green transition, the industry needs to take action:
- Renewable energy:
Data centers must be powered by solar and wind to cut carbon emissions. Companies like Google, Microsoft, and Amazon are already investing in renewable energy to make data centers more self-sufficient and energy-efficient. - More energy-efficient models:
Emerging technologies such as quantum computing and improved algorithms could reduce energy consumption during training and inference of AI models. - Hardware recycling:
Designing circuit boards and hardware for easier disassembly and material recovery can help reduce the need for new raw materials. - International regulations:
Global frameworks are needed to ensure that mineral extraction happens under responsible conditions—for both workers and the environment.
Conclusion
AI is a double-edged sword in the fight against climate change. It has the power to cut emissions and drive sustainable innovation, but it can also deepen environmental issues if its resource demands and industry practices aren’t responsibly managed.
The DeepSeek example shows that it is possible to develop AI that is both cost-efficient and less resource-intensive—offering hope for a more sustainable future. At the same time, it reminds us of the need for transparency and accountability in an industry still struggling to balance rapid growth with environmental impact.
The question remains: how do we balance the potential of technology with its costs?
By focusing on innovation, promoting energy efficiency, and making responsible choices, we can shape a future where AI becomes part of the solution—not the problem.
Sources:
- Strubell, E. m.fl. (2019). Energy and Policy Considerations for Deep Learning in NLP.
- Microsoft Sustainability Report (2022)
- UNEP.org: Rapport om bærekraftig gruvedrift
- IPCC: Klimaendringer og teknologiske løsninger
- NRK: “Slik sluker KI-tjenester vannressurser” – Shaolei Ren om vannforbruk knyttet til AI-tjenester.
- SINTEF: “Vannressursenes rolle i teknologisk utvikling”.
- Business Insider: “What is DeepSeek? Get to know the Chinese startup that shocked the AI industry”
