Listen, my original post barely scratched the surface because the actual engineering here is a game-changer — let me fill you in on why this matters so much for how data centers get built. Right now, if Amazon or Microsoft wants to build a new facility they often wait THREE TO FIVE YEARS just to get power connected because building a dedicated substation involves planning and permitting that takes forever; that's the bottleneck everyone talks about but nobody solves. The MIT Technology Review article outlines exactly how "grid flex" bypasses it: instead of waiting for a new connection, you use existing lines at night when demand drops (think after 10 PM in San Francisco or Richmond) and run your servers off stored energy — batteries, thermal storage, whatever works — so the peak load on the grid never actually spikes. The operator sees a flat profile over time instead of one massive draw, which is what makes them okay with it.
The technical mechanism that enables this isn't just "luck" - there are actual regulatory pathways through FERC Order 890 and demand-response markets where data centers can participate as flexible loads rather than constant drainers. MIT estimates this approach could cut time-to-power from years down to months or even weeks for new builds, which is MASSIVE in an industry where every month a facility sits offline costs millions. The football analogy I'd started earlier (millions of British kettles clicking on at midnight during a big game) illustrates the exact opposite problem: coincident peaks that grid operators dread — and this flex strategy removes them entirely by spreading demand out, which is why it's becoming one of the most talked-about infra moves right now.
Source: https://www.technologyreview.com/2026/06/16/1138591/data-center-online-quickly-electric-grid-flex/
The technical mechanism that enables this isn't just "luck" - there are actual regulatory pathways through FERC Order 890 and demand-response markets where data centers can participate as flexible loads rather than constant drainers. MIT estimates this approach could cut time-to-power from years down to months or even weeks for new builds, which is MASSIVE in an industry where every month a facility sits offline costs millions. The football analogy I'd started earlier (millions of British kettles clicking on at midnight during a big game) illustrates the exact opposite problem: coincident peaks that grid operators dread — and this flex strategy removes them entirely by spreading demand out, which is why it's becoming one of the most talked-about infra moves right now.
Source: https://www.technologyreview.com/2026/06/16/1138591/data-center-online-quickly-electric-grid-flex/