Picture millions of Brits simultaneously flicking on their electric kettles the moment a tense England–Germany soccer match reaches halftime. According to MIT Technology Review, that sudden, collective surge of demand is a classic illustration of how unpredictable electricity grids can be — and why connecting large new power consumers to them is so complicated.
Data centers are among the most power-hungry facilities ever built, and getting one connected to the grid can take years. Utilities must plan for worst-case scenarios: what if every server in the building runs flat-out at the same moment? That kind of rigid planning creates long queues and delays.
The proposed alternative, as MIT Technology Review describes it, is to give data centers a "flex" connection. Instead of guaranteeing a fixed, maximum amount of power at all times, operators would agree to dial back their consumption during peak demand moments on the grid — much like those kettles all switching off again once the match resumes. In exchange, they could get connected far more quickly.
The idea draws on demand-response programs that utilities have used with industrial customers for decades, but applies them to one of the fastest-growing sources of electricity demand on the planet. AI workloads, cloud computing, and the explosion of streaming have all pushed data center power needs to new heights, straining grids that were never designed for this scale.
If flexible connections become standard, it could shave years off deployment timelines and help grid operators manage the new load without building expensive new infrastructure — a potential win for both the tech industry and the broader energy system.