However, those dense ranks of cabinets eat up gigawatts of power and LLM training produces spikes in that appetite for electricity.<\/p>\n
These spikes are equivalent to thousands of homes switching kettles on and off in unison every few seconds.<\/p>\n
This type of irregular demand on a local grid needs to be carefully managed.<\/p>\n
Daniel Bizo of data centre engineering consultancy The Uptime Institute analyses data centres for a living. <\/p>\n
“Normal data centres are a steady hum in the background compared to the demand an AI workload makes on the grid.”<\/p>\n
Just like those synchronised kettles sudden AI surges present what Mr Bizo calls a singluar problem.<\/p>\n
“The singular workload at this scale is unheard of,” says Mr Bizo, “it’s such an extreme engineering challenge, it’s like the Apollo programme.”<\/p>\n
Data centre operators are getting around the energy problem in various ways.<\/p>\n
Speaking to the BBC earlier this month, Nvidia CEO Jensen Huang said that in the UK in the short term he was hoping that more gas turbines could be used “off the grid so we don’t burden people on the grid”.<\/p>\n
He said AI itself would design better gas turbines, solar panels, wind turbines and fusion energy to produce more cost effective sustainable energy.<\/p>\n
Microsoft is investing billions of dollars in energy projects, including a deal with Constellation Energy that will see nuclear power produced again<\/a> on Three Mile Island.<\/p>\n Google, owned by Alphabet, is also investing in nuclear power, external<\/a> as part of a strategy to run on carbon-free energy by 2030, external<\/a>.<\/p>\n