‘We cannot afford the electricity bill’: Sridhar Vembu warns AI could overload India’s power grid
What if the future of AI meant rolling blackouts for your home? Zoho founder Sridhar Vembu says that’s not science fiction, it’s a looming crisis as data-hungry AI models drive power bills sky-high and stretch electric grids to the brink.
In a stark warning on X, Vembu pointed to a 60% jump in electricity costs since 2023 in Athens, Georgia—blaming the spike on AI-driven data centers.
“Even if we could afford all the GPUs (not!), we cannot afford the electricity bill,” he wrote. “We cannot afford to hurt households and factories.”
Vembu called today’s AI “extraordinarily energy inefficient” and said India must rethink how it builds digital infrastructure.
“We need vastly more energy efficient AI,” he said, calling for a fundamental shift in how AI is computed.
India’s data center capacity is projected to balloon from 1.2 GW in 2024 to as much as 5 GW by 2030.
AI workloads could consume 40–50 terawatt-hours annually by then—straining a power system that’s already under pressure to meet surging industrial and residential demand.
Even as India ramps up its total generation capacity to 777 GW, experts warn that any shortfall in renewable energy targets could create regional deficits, especially in AI-heavy zones. If clean energy growth slips, the nation could face a 6 TWh annual gap—with the risk of blackouts during heatwaves or high AI activity.
Local grids may not be equipped to handle the concentrated demand of data centers, raising the specter of regional outages that hit homes and factories alike.
