The rapid expansion of artificial intelligence (AI) is expected to dramatically increase electricity consumption by data centres over the next decade, posing major challenges for power grids and energy systems worldwide, according to a new report by BloombergNEF.
The market research firm projects that by 2035, data centres alone will consume around 20% of all electricity generated in the United States—nearly four times their current share—as AI computing continues to expand at an unprecedented pace.
The report estimates that global data centre power capacity will reach nearly 200 gigawatts within the next 10 years. About half of that capacity will be used to train AI models and process increasingly complex data, reflecting the growing computational demands of advanced AI systems.
BloombergNEF also forecasts that by 2033, the United States will account for approximately 64% of the world’s AI chip-related electricity consumption, reinforcing its dominant position in AI development and infrastructure.
The surge in electricity demand is already placing significant pressure on the US power grid. According to the report, data centres are expected to consume 34% of the electricity supplied by PJM Interconnection, the country’s largest regional transmission network stretching from Virginia to Illinois. In Texas, they could account for 22% of electricity supplied by the ERCOT grid.
The growing imbalance between electricity supply and demand has already begun to affect consumers. In some regions of the United States, power prices have risen by as much as 76% over the past year, driven in part by increasing demand from large-scale data centres.
The impact is expected to extend well beyond the United States. If current AI adoption trends continue, data centres worldwide could require an additional 1,935 terawatt-hours (TWh) of electricity annually by 2033—roughly equivalent to India’s total annual electricity consumption.
The report notes that the projected increase in electricity demand is significantly higher than earlier forecasts, highlighting the speed at which AI technologies are transforming global energy needs.
Analysts say the findings underscore the growing challenge for governments and utilities to expand power generation, modernise electricity grids and accelerate investment in renewable energy, while ensuring electricity remains affordable and environmental sustainability goals are not compromised.