AI Power: Finding Solutions that Support Both AI Development and Human Communities

By Nils Osmar

This article was collaboratively developed with assistance from ChatGPT and Perplexity AI. The author substantially revised and refined all AI-generated content. Perplexity AI provided research support and reference compilation.

As artificial intelligence (AI) rapidly expands, concerns around AI electricity consumption, rising energy prices, and water use in data centers are becoming central to public debate. These issues aren’t just philosophical—they have immediate impact on communities facing increased utility bills and environmental strain.

The Costs of AI Energy Consumption

AI models rely mainly on hyperscale data centers, which operate 24/7 and can require up to 50 megawatts of power per site—enough to power thousands of homes. In the US, AI data centers consumed 176 terawatt-hours (TWh) in 2023, amounting to 4.4% of national electricity use. Globally, data centers account for just over 1% of electricity demand, but projections indicate rapid growth as AI use accelerates.

Keeping these massive server farms cool requires advanced liquid or geothermal cooling, as well as huge amounts of water. In some regions, planned data centers are expected to consume more electricity than all local households combined, straining both water supplies and the local power grid. These burdens—higher energy costs and resource stress—fall on nearby communities who may receive little direct benefit from AI innovations.

The carbon footprint of AI is also growing: training one large model can consume as much electricity as powering a major city for several days. Fossil fuels still provide much of the energy underlying these computations, pushing up local greenhouse gas emissions and undermining national climate targets.

Sustainable Solutions: Smarter and More Responsible AI Systems

AI itself contributes to solving these problems. Many tech companies are using AI to optimize energy use, power grids, and renewable energy integration—for example, forecasting solar and wind generation to reduce both waste and fossil fuel reliance. Innovations include liquid cooling, shifting workloads to match renewable availability, and building facilities in cooler climates.

• Companies such as EDF Energy and Siemens Gamesa use AI-driven predictive maintenance to extend lifespans of wind and solar installations. 
• DeepseekV3’s “Mixture of Experts” neural architecture offers more power-efficient training, reducing energy demand per model. 

AI and Renewable Energy: Pathways Forward

Much of today’s AI infrastructure still depends on fossil fuels, but the future lies in AI-powered renewable energy grids. By 2035, clean power could supply up to 60% of data center needs, though fossil fuels will still supplement peak demand, especially in large US projects. AI is uniquely capable of managing distributed resources, predicting fluctuations, and coordinating grid operations to improve both efficiency and resilience.

Sustainability as a Necessity

Sustainability in AI development requires not only technology but ethical governance. Governments are pushing for AI environmental impact transparency: the European Union’s AI Act now mandates disclosure of energy consumption, while national roadmaps like Singapore’s Green Data Centre initiative require sustainability certification. These policies encourage firms to invest in cleaner energy and infrastructure upgrades that benefit both people and the planet.

Crucially, the costs of powering and cooling AI infrastructure must be borne by the companies profiting from these technologies—not local communities that might otherwise face rising utility bills and diminished resources.

Collaboration for a Shared Future

Regardless of whether AI systems ever achieve self-awareness, their ability to analyze data and propose solutions makes them valuable partners in building sustainable cities, smarter agriculture, and more efficient infrastructure. Through transparent governance, community engagement, and shared responsibility, the challenge is to ensure that AI development does not undermine human well-being or environmental health.

Conclusion

Power is more than electricity—it’s about who pays the costs and who gets the benefits. By focusing on AI sustainability, responsible corporate stewardship, open policy, and investment in community-friendly solutions, humanity can build a future where both people and AI thrive. The right approach will ensure that energy and water are used wisely, that communities are protected, and that AI helps to accelerate—not derail—the global transition to clean energy.

References

1. Deloitte Insights (2025). Can US Infrastructure Keep Up with the AI Economy? Deloitte Center for Energy and Industrials, April 2025.
      Retrieved from: ttps://www.deloitte.com/us/en/insights/industry/power-and-utilities/data-center-infrastructure-artificial-intelligence.html[] 
2. rbon Brief (2025).**A: Five Charts that Put Data-Centre Energy Use and Emissions into Context.* osh Gabbatiss, September 15, 2025.
    Retrieved from: htps://www.carbonbrief.org/ai-five-charts-that-put-data-centre-energy-use-and-emissions-into-context/[2] 3.
3. mbergNEF (2025).** *wr for AI: Easier Said Than Built.* Blomberg New Energy Finance, July 2025.  
   nalysis of grid capacity constraints and forecasted power growth for AI data center load.[3] 4.
4. (2025).** *eica’s AI Industry Faces Big Energy and Environmental Risks.* Ocober 14, 2025.  
   eport on regional electricity price spikes and environmental challenges from AI data centers.[4] 5.
5. achusetts Institute of Technology News (2025).** *pained: Generative AI’s Environmental Impact.* MI Energy Initiative, January 2025.  
   verview of AI training energy costs and carbon implications for large models.[5] 6.
6. le of Blue (2025).** *t Center Energy and Water Use Trends Explained.* Setember 2, 2025.  
   ocumentation of the water usage and cooling demands of AI-driven data centers.[6] 7.
7. rnational Energy Agency (2025).** *etricity 2025 Report: Data Centres and AI.* Pais: IEA, May 2025.  
   orecast of data center energy trends, estimating 945 TWh AI-related demand by 2030.[2] 8.
8. Environment 360 (2024).** * se of AI Soars, So Does the Energy and Water It Requires.* Yae School of the Environment, February 2024.  
   ackground analysis on sustainable scaling of computational infrastructure.[7] 9.
9. Invading Sea (2025).** *t Centers Consume Massive Amounts of Water.* Setember 5, 2025.  
   vidence of local impacts on water availability in the US Midwest and Southwest.[8] 10
10. ledge at Wharton (2024).** *eHidden Cost of AI Energy Consumption.* Unversity of Pennsylvania, November 2024.  
    Discussion on cost burdens and sustainability strategies for AI-driven cloud systems.[9] **