Texas freezes new data-center grid connections while a 474 GW queue waits on audit.
What happened: Governor Greg Abbott directed the Public Utility Commission of Texas and ERCOT to complete a comprehensive verification and audit of data centers advancing through the interconnection process before any project moves forward. Projects that fail PUCT/ERCOT requirements are to be denied connection to the Texas grid. ERCOT is considering more than 474 gigawatts of connection requests — more than five times Texas’ record peak demand — with roughly 90% of those requests coming from data centers. Ars Technica reports the order as a practical moratorium on new data-center grid connections until developers provide more impact information. The same infrastructure boom is hitting physical-labor bottlenecks: NBC News reports U.S. data-center builds are short tens of thousands of fiber-optic installers needed to connect facilities to the internet.
Why it matters: This is a hard reliability and siting signal, not a global electricity-share story. The measurable record is whether the audit culls speculative queue entries, how long the freeze lasts, who pays for upgrades, and whether construction labor shortages delay projects even when power is available.
Community opposition blocked or delayed about $130B of data-center projects in Q1 2026.
What happened: A Data Center Watch study shared with NBC News finds opponents blocked or delayed at least 75 data-center projects nationwide worth about $130 billion from January through March 2026 — the most in any three-month period since tracking began in 2023, and roughly matching the total blocked or delayed across all of 2025. Authors describe a structural shift: communities have internalized an opposition playbook, legislatures added regulatory uncertainty, and active opposition groups more than doubled to 833 across 49 states. Local fights center on energy use, environment, noise, and neighborhood impacts rather than abstract AI hype.
Why it matters: Permitting and politics are now a first-order constraint on AI infrastructure, not a side story. The measurable record is project cancellations, delay length, state moratoriums, and whether ratepayer/local-burden rules actually change who hosts new load.
Etsy cuts ~12% of staff after a revenue beat — and says the restructure is not an AI replacement story.
What happened: Etsy told shareholders it is laying off about 220 people, roughly 12% of its workforce, while reporting year-on-year Q2 revenue growth above 6% (excluding Depop) and a swing to a $46.7 million net loss. In a memo made public the same day, CEO Kruti Patel Goyal said neither cost cuts nor artificial intelligence spurred the move, framing it as organizational restructuring toward fewer silos and flatter, faster teams. She also said AI is changing how people work and how products get built, while insisting the future depends on combining people with technology rather than replacing one with the other. Coverage notes the cuts fall heavily on product and engineering teams.
Why it matters: This is a live case of the AI-era ambiguity in corporate headcount: restructuring language, AI-adjacent product rebuilding, and an explicit denial that AI caused the cuts. The measurable record is role mix, hiring after the reorg, and whether similar “not AI, but AI is changing work” memos become the default template.
Nearly 1,900 U.S. public schools sit within a mile of a data center — with lower nearby math scores flagged.
What happened: An exclusive The 74 analysis of Brown University research finds 1,878 U.S. public schools within a mile of at least one existing or approved data center. California alone accounts for 602 schools (about 32% of the national total). The Brown team’s forthcoming paper reports lower math scores among students attending schools closer to data centers. Local conflicts — from Coweta County, Georgia’s Project Sail near Arnco-Sargent Elementary to Palm Beach County’s rejection of a project near Saddle View Elementary — center on noise, diesel generators, heat-island effects, traffic, and water quality. Industry groups say impacts can be mitigated with buffers and design; school board members and residents say facilities should not sit next to classrooms.
Why it matters: AI infrastructure is becoming a child-health and learning-environment issue, not only a grid or climate issue. Treat the math-score finding carefully: it is forthcoming research shared via reporting, not yet a peer-reviewed causal proof. The measurable record is school proximity counts, noise/air monitoring near campuses, and whether siting rules create school buffers.
Schools turn to AI to score student work beyond A–F grades after a mastery-transcript acquisition.
What happened: The 74 reports that New York-based Legend.org bought the nonprofit Mastery Transcript Consortium and is building AI tools to help teachers rate academic and “soft” skills at schools that skip traditional A–F report cards. The platform is aimed at about 400 public and private consortium schools. Teachers can upload digital files or photos of handwritten work; AI evaluates against state, school, or teacher standards and is also being developed to score recorded presentations and other evidence of communication, collaboration, and critical thinking.
Why it matters: This is a live shift from AI-as-cheating risk to AI-as-assessment infrastructure. The measurable record is whether teachers remain accountable for final judgments, how soft-skill rubrics are validated, and whether students can contest machine-assisted scores.
