Artificial intelligence (AI) is reshaping business and society, but its rapid growth comes with a massive energy cost. Here is a look at the progress being made and the profound challenges that remain.

The Good News: Remarkable Efficiency Gains

In August 2025, Google reported significant improvements in its Gemini platform. A single text prompt now:

  • Uses 0.24 watt-hours of electricity.

  • Emits just 0.03 grams of CO₂e.

  • Consumes 0.26 millilitres of water (about five drops).

Over the past year, the median prompt’s energy use fell 33-fold and its carbon footprint dropped 44-fold. These improvements reflect gains across hardware, software and data centres. Researchers caution, however, that such measurements may not capture all indirect impacts.

The Sobering Reality: Demand is Outpacing Savings

Despite these gains, total consumption is climbing rapidly.

  • U.S. data centres consumed about 4.4 per cent of national electricity in 2023, and are projected to reach 6.7 to 12 per cent by 2028.

  • Computational performance per watt has improved by up to 150 per cent annually since 2019, but these efficiency gains are not enough to offset the sheer scale of new demand.

The Ultimate Bottleneck: Energy Itself

Efficiency alone cannot solve the challenge. Industry leaders increasingly frame electricity as the primary bottleneck for scaling AI. During a U.S. Senate hearing in May 2025, OpenAI’s CEO highlighted energy infrastructure as a key limit to AI growth.

Various analyses suggest the United States may need 50 to 90 gigawatts of additional generating capacity in the coming years to decade to meet rising data-centre demand. For context, some estimates indicate utilities could require around $50 billion in new power generation capacity by 2030 for data centres alone.

The Path Forward

Efficiency gains are critical, but they are not a silver bullet. Sustaining AI’s growth requires a three-pronged approach:

  1. Radical Transparency: Clear, standardised reporting on energy and water use.

  2. Continuous Innovation: Ongoing advances in hardware, software and cooling systems.

  3. Clean Energy at Scale: Faster deployment of carbon-free power, including 24/7 matching commitments, where renewable energy procurement aligns with actual consumption patterns in real time.

Conclusion

AI’s growth brings both opportunity and responsibility. Efficiency gains are encouraging, but absolute consumption will continue to rise unless they are paired with clean energy and grid expansion. As of August 2025, the imperative is clear: scale innovation and sustainability in tandem.

What are your thoughts on balancing AI innovation with sustainability? Perspectives from energy experts and AI leaders are welcome in the comments.

#AI #ArtificialIntelligence #Sustainability #EnergyEfficiency #ClimateTech #DataCenters #TechInnovation #CleanEnergy #CarbonNeutral #GreenTech #FutureOfWork #DigitalTransformation #SmartEnergy #EnergyTransition #RenewableEnergy #GridInnovation #CloudComputing #ICT #TechLeadership #Cybersecurity #EmergingTech #Innovation #ClimateAction #BusinessStrategy #ResilientInfrastructure #SmartInfrastructure #EnergyPolicy #GlobalTrends #ResponsibleAI #TechnologyStrategy #ITInfrastructure #SustainableGrowth #GreenEnergy #SustainableTech #EnergyFuture