AI and Environment: Carbon Cost of Artificial Intelligence
Why in the news
As AI investment booms, concern is growing over its environmental footprint and the lack of sustainability rules.
Key facts
- AI is valued at about $200 billion now and may add $15.7 trillion to the world economy by 2030.
- Major investments: the US Stargate Project ($500 billion), a Reliance-Nvidia tie-up for the world’s largest data centre in India, and an Indian LLM project.
Environmental impact
| Issue | Detail |
|---|---|
| Data centres | About 1% of global CO₂; expected to double by 2026 (IEA) |
| Generative AI | Needs 100x more computing power |
| Training | GPT-3 emitted 552 tonnes of CO₂ |
| E-waste | Demand for GPUs and TPUs adds to waste |
Policy gaps
- ITU and COP29 stressed sustainable AI, yet 190+ countries adopted only non-binding ethics.
- EU and US rules largely miss sustainability; national strategies overlook private-sector roles.
Way forward
- Run data centres on renewables; DeepMind used AI to improve wind forecasting.
- Use efficient hardware, domain-specific models and pre-trained models.
- Measure and publish emissions via standard carbon tracking.
Exam angle
- Source of emission estimate: International Energy Agency.
- Related terms: Stargate Project, LLM, e-waste.