Artificial intelligence is driving a huge rise in electricity demand but the same technology could also help reduce the amount of renewable power wasted through grid curtailment.
The global AI market is forecast to grow from $190 billion (£142bn) in 2023 to almost $5 trillion (£3.74tn) by 2033, while worldwide spending on the technology could exceed $2.5 trillion (£1.87tn) in 2026 alone.
That expansion will require vast amounts of electricity, particularly for data centres and the training of increasingly powerful AI models.
Data centres were expected to consume around 450TWh in 2025, which would place them 11th globally if ranked as a country by electricity use.
Demand could reach 1,000TWh by 2030 and rise to 1,300TWh in later years, exceeding the consumption of some entire economies. Training GPT-4 alone is estimated to have used up to 70GWh of electricity over 100 days, around 50 times more than GPT-3.
Markus Voigt, Executive Chairman of German renewables firm the aream Group, said: “Renewable energies therefore play a key role in the widespread adoption of this technology of the future.”
However, aream’s own renewable energy portfolio shows how much clean electricity is already being lost because grids cannot always absorb the power available.
Its German solar plants were curtailed by around 1,300MWh during June, a prolonged grid outage and inverter failures caused by high temperatures led to a further loss of around 300MWh.
Spanish solar assets were curtailed by around 3,350MWh, wind farms also lost around 850MWh through curtailment, alongside further disruption caused by equipment faults and grid works.
AI could potentially help address the problem by forecasting renewable generation and electricity demand more accurately, improving battery dispatch and helping network operators balance local grids in real time.
Better data centre flexibility could also allow operators to shift some computing loads to periods when wind and solar output is high, reducing pressure during peak demand and avoiding the shutdown of renewable assets.
Wind and solar are already expected to meet almost half of the additional electricity demand from data centres by 2030, supported by long-term power purchase agreements between technology companies and generators.
Copyright © 2026 Energy Live News LtdELN