AI’s Impact on Power Supply: A Growing Concern

AI’s impact on power supply poses a significant issue, as Zoho founder Sridhar Vembu highlights that potential rolling blackouts could become a reality for homes. This concern stems from the soaring electricity costs driven by data-intensive AI models that are straining electric grids to their limits.

Electricity Costs on the Rise

In a candid statement on X, Vembu pointed out a staggering 60% rise in electricity costs since 2023 in Athens, Georgia, attributing the increase to the demands of AI-driven data centres. He remarked that despite the possibility of acquiring more GPUs, the electricity expenses are becoming unmanageable. The well-being of households and factories is at risk due to these escalating costs.

Energy Inefficiency of Current AI

Vembu described contemporary AI as “extraordinarily energy inefficient,” urging India to reconsider its approach to digital infrastructure. He emphasised the necessity for more energy-efficient AI models, advocating for a fundamental change in the computation of AI.

Future Projections for India’s Data Centers

India’s data centre capacity is anticipated to surge from 1.2 GW in 2024 to possibly 5 GW by the year 2030. Projections suggest that AI workloads may consume between 40 and 50 terawatt-hours every year by that time, which could burden a power system that is already striving to satisfy increasing industrial and residential energy needs.

Challenges Ahead for Renewable Energy

As India increases its total generation capacity to 777 GW, experts caution that any setbacks in renewable energy targets could lead to regional deficits, particularly in areas heavily reliant on AI. If the growth of clean energy falters, the country might encounter a shortfall of 6 TWh annually, elevating the risk of blackouts during periods of extreme heat or elevated AI usage.

Risks to Local Grids

Local grids may struggle to support the concentrated energy demands of data centres, raising concerns about regional outages that would affect both residential and industrial sectors.