Agreed on all counts. The post isn't arguing against AI infrastructure, just for more transparency about the current costs and more intentionality in how we use it while that clean baseload buildout happens.
The carbon-aware computing angle is actually in the full piece. It's one of the more promising near-term approaches since it doesn't require waiting for new plants to come online.
The irony of asking AI to solve its own energy problem isn't lost on me. Though to be fair, this post did use Claude for research, which I estimated at roughly 26 watt-hours and 8 gallons of water. Figured I should practice what I preach and disclose that.
The carbon-aware computing angle is actually in the full piece. It's one of the more promising near-term approaches since it doesn't require waiting for new plants to come online.
The irony of asking AI to solve its own energy problem isn't lost on me. Though to be fair, this post did use Claude for research, which I estimated at roughly 26 watt-hours and 8 gallons of water. Figured I should practice what I preach and disclose that.