IEA: data-centre electricity jumped 17% in 2025 — bottlenecks and offtake now define the near-term AI power story.
Observed-2025 layer — not earlier IEA 2030 TWh paths; not PJM ride-through.
Read today’s coverageFor people trying to make sense of AI
Track AI's footprint without the spin.
AI FootprintTM tracks how AI is changing work, infrastructure, policy, health, science, education, and culture. The front page carries the most important headlines; the category pages carry today's full list by beat.
Updated September 19, 2026Latest edition September 19, 2026Front page editor's picksCategories full daily lists
Headline articles · September 19
The strongest stories from the latest edition.
Minneapolis Fed: automating records work can raise average wages — and still hurt the people whose skill was the records.
Within-occupation transformation — not Boston Fed CPP 26-8; not a layoff print.
Read today’s coverageNIST puts an AI agent on the National Vulnerability Database — modernization comments due 13 October.
NVD agent workflow — not Article 50 labels; not FDA 510(k) denial.
Read today’s coverageFirst-trimester ML beat existing early pregnancy risk tools across more than half a million pregnancies — Chile still misfit.
Retrospective risk tool — not a bedside RCT; not Nature Medicine evidence comment.
Read today’s coverageDaily beats
Each page has today's items for that lane.
Positive stories are included where they belong. A medical breakthrough goes under Health & Science. A job-gain story goes under Jobs. An energy-efficiency story goes under Environment.
01
Jobs
Layoffs, hiring, wages, productivity, automation, job creation, and worker leverage.
02Environment
Energy demand, water use, emissions, data centers, grids, chips, and local siting conflicts.
03Policy
Regulation, safety, courts, procurement, enforcement, and public-sector governance.
04Health & Science
Medicine, biology, clinical AI, scientific discovery, and research infrastructure.
05Education & Culture
Schools, children, media literacy, creative work, synthetic media, and public trust.
Editorial standard
Useful, not theatrical.
1
Today means today.
Current pages carry the latest publishing window; older material goes to the archive.
2
One story, right lane.
Benefits are not siloed. Good news is filed under the category it affects.
3
Source dates matter.
Edition dates and story dates are kept separate.
4
Keep it human.
The point is to help readers see what changed and why it matters.
Origin
This started with buffalo cauliflower.
Frank made a funny AI image during a work trip. A coworker said it was wasteful and bad for the environment. Instead of dismissing that concern, he wanted a better answer: what does one AI use actually cost, and how should people think about all the bigger uses happening every day?
Read the origin story