Editorial 2: Redrawing the not-so-pretty energy footprint of AI
Context
Small modular nuclear reactors might be the solution to power the rapidly growing AI and data infrastructure.
Introduction
Generative AI has made it much easier to create art and finish tasks faster and with less effort. For example, ChatGPT-4o can make a Studio Ghibli-style portrait in just a few seconds using a simple prompt. But this convenience uses a lot of energy, which people often don’t notice — sometimes even causing GPUs to overheat or melt. As AI tools get more powerful, their harmful impact on the environment will grow, making them less sustainable. So, how can we build AI in a way that’s better for the planet? Could using nuclear energy, especially Small Modular Reactors (SMRs), be a good alternative?
The Environmental Impact and Energy Challenges of AI Adoption
- Hidden Cost of AI Use: AI tools are not free — every time someone uses ChatGPT or any other AI application, a data centre somewhere consumes a large amount of electricity, much of it generated from fossil fuels.
- Hardware Strain and Overload: OpenAI CEO Sam Altman highlighted the strain on hardware, tweeting: “It’s super fun seeing people love images in ChatGPT, but our GPUs are melting.”
- Rising Global Electricity Demand: By 2030, projections suggest that data centres could consume up to 10% of the world’s total electricity.
- India’s Present Readiness: India currently has sufficient capacity to generate electricity for its domestic AI needs, but with increasing adoption and ambitions, proactive planning is necessary.
- Carbon Emissions from AI Training: Training a single AI model — whether for conversation (like ChatGPT) or image generation (like Midjourney) — can produce as much CO₂ as five cars running across their lifetimes.
- Continuous Power Demand Post-Deployment: Even after deployment, AI tools continue to consume immense energy from data centres to serve millions of users globally.
- Sustainability Challenge: This level of energy consumption is becoming increasingly unsustainable as AI adoption grows worldwide.
- Need for Energy Transparency: AI companies should disclose three key details: how much energy they use, where it comes from, and what measures they are taking to reduce consumption.
- Regulatory Framework for AI Energy Use: Just like data privacy regulations, there should be rules requiring companies to report their environmental impact and energy usage clearly.
- Insight and Innovation for Sustainability: Energy usage data can help identify high-consumption areas, promote targeted research, and drive the development of more sustainable AI technologies.
Advantages of SMRs
|
Aspect
|
Details
|
|
Core Idea
|
A shift in focus is needed toward the source of energy powering technological growth — particularly nuclear energy, with emphasis on Small Modular Reactors (SMRs).
|
|
Urgency
|
The AI boom is expanding rapidly, and the current energy infrastructure is not equipped to keep up with its power demands.
|
|
SMRs vs Traditional Nuclear Plants
|
SMRs are compact, scalable, and demand less land, water, and infrastructure compared to large nuclear power plants.
|
|
Deployment Advantage
|
SMRs can be installed closer to data centres and other high-energy demand sites, providing consistent and reliable power.
|
|
Carbon Impact
|
SMRs provide 24×7, zero-carbon, baseload electricity, making them a strong alternative to intermittent renewables like solar and wind.
|
|
Construction Benefits
|
Their modular design allows for faster construction, lower costs, and quicker deploymentcompared to traditional plants.
|
|
Safety Features
|
SMRs have enhanced passive safety systems that use natural cooling processes to prevent overheating or accidents.
|
|
Public Acceptance
|
Safer design and smaller size make SMRs more socially acceptable, especially in areas resistant to large-scale nuclear projects.
|
|
Geographical Flexibility
|
SMRs can operate in a wide range of environments, from urban to remote areas, supporting energy decentralisation.
|
|
Grid Resilience
|
By producing energy closer to consumption points, SMRs reduce transmission losses and improve overall grid resilience.
|
Some of the challenges
- Challenges in SMR Adoption: The adoption of Small Modular Reactors (SMRs) is not without challenges. Significant policy shifts will be necessary to create a robust regulatory framework that addresses safety, waste management, and public perception.
- Upfront Investment and Cost Competitiveness: There is also the matter of substantial upfront investment, as the technology is still maturing and may face issues with cost competitiveness when compared to established energy sources.
- Coordinating with Renewable Energy Initiatives: Coordinating the deployment of SMRs with existing renewable energy projects will require careful planning to maximize synergies while minimizing redundancy.
- Electricity Costs in India: Despite these challenges, the cost of electricity from SMRs in India is predicted to fall from ₹10.3 to ₹5 per kWhonce the reactors are operational, which is less than the average electricity cost.
Conclusion
In conclusion, a public-private partnership model offers a realistic solution to the challenges of sustainable AI development. By leveraging the strengths of both the public and private sectors, this model can enable the efficient development of SMRs alongside other forms of renewable energy to support advancements in AI.