Every article we've published — newest first.
AMD pairs Microsoft-hosted autonomous chip design with its Taalas buy, cutting debug time 25-40% and tape-out to about two months for custom inference chips.
We break down Amazon's on‑site gas plant, grid‑connection plans for its 8,000‑acre AI campus and the PJM outlook shaping large‑load power needs.
CoreWeave is building its first APAC data centers in Indonesia — 360 MW of AI-focused capacity, fully owned and operated, coming online in 2028.
Southeast Asia's AI data center pipeline hit a record 26.5 GW in H1 2026, with 31 sites over 100 MW planned, as Nvidia GB300 chip access draws scrutiny.
How ultracapacitors, natural‑gas generation and utility‑scale renewables are being combined to power AI workloads.
Explore how ultra‑compact AI packages drive high‑density AI cooling, from direct‑to‑chip liquid solutions to integrated thermal architectures.
We examine the Cherokee Nation's ban on hyperscale AI data centers, community push‑back, and the legal obstacles shaping U.S. data‑center projects.
Explore the scarcity of high‑density AI data‑center space, power‑density trends, construction costs, and design rules for mega‑AI campuses.
Nvidia's GB300 NVL72 draws up to 142kW a rack (~15x a typical enterprise rack), runs $3.7-4.0M, and just went live at the Naval Postgraduate School.
OpenAI and Oracle scrapped the 2GW Abilene expansion, leaving 1.2GW built. The $3B Nvidia power deal, $90B in new debt, and who could claim the spare capacity next.
How Texas’s 474 GW data‑center surge threatens grid reliability, the audit moratorium, and new co‑location rules for AI workloads.