Infrastructure, analyzed daily.
AI infrastructure, cybersecurity, power and cooling — the announcements that matter, with the analysis and the unanswered questions the press releases leave out.
NVIDIA Blackwell Tops the First Agentic AI Infrastructure Benchmark
NVIDIA reports its Blackwell platform leads the first agentic AI infrastructure benchmark, a new test of multi-step, tool-using inference workloads. We assess what the vendor-reported result covers, what…
Amazon’s 7x Water Claim Rests on Hotter Servers, a Test for AI-Era Cooling
Amazon says its data centers used 0.12 liters of water per kilowatt-hour in 2025, over 7x better than a 0.84 industry average. The gain rests on free-air cooling…
Louisville’s 500,000 Sq Ft Cap Shows Local Zoning Now Gates AI Data Center Scale
Louisville proposed banning data centers over 500,000 square feet and Lexington paused data center applications through Oct. 31. Kentucky’s largest cities are showing that the size of AI…
Texas Governor Calls for Regulators to Rein In Data Centers
Texas’ governor has called for a regulatory clampdown on data centers, a notable policy turn in America’s fastest-growing data center market. We examine what is known so far,…
Gartner: Data Center Electricity Use to Grow 26% in 2026
Gartner forecasts data-center electricity consumption will grow 26% in 2026, an acceleration driven by AI workloads that utilities must now plan around. We analyze what the projection means…
AMD Says Instinct MI355X Sets a New Bar for DeepSeek Inference
AMD claims its Instinct MI355X GPU sets a new performance bar for DeepSeek inference, a direct challenge to Nvidia in the fast-growing market for serving AI models. We…
Ohio Would Halve a $1.6B Data Center Break. The Harder AI Fight Is Who Pays for the Grid
Ohio Sen. Brian Chavez’s data center bill would halve a sales tax break that reached about $1.6 billion last year and add a data center utility rate class.…
FERC Approves PJM’s Temporary Fast-Track for Large Capacity Projects
FERC approved PJM’s temporary fast-track process for large capacity projects, letting select power resources move through the grid operator’s queue faster. We examine what the decision means for…
d-Matrix Corsair Hits Full Production: A Challenger to the AI Inference Status Quo
d-Matrix Corsair, an AI inference accelerator built on digital in-memory compute, has entered full production, the startup says, citing customer demand. We examine what the milestone means for…