Topic

AI Infrastructure

GPU clusters, training campuses, and the hardware race behind AI.

Where the Coal Plant Stood: AWS Files a Ten-Year, 2,200-Acre Data Center Plan at Homer City

Amazon's development arm has filed a master plan for the Homer City Energy Campus in Indiana County, Pennsylvania: dozens of two-story data centers on six sites straddling two townships, built in phases from December 2026 to 2036 and fed by a 4.5 GW gas plant rising where Pennsylvania's largest coal plant stood until 2023. More than 100 residents packed the county courthouse and the planning commission tabled it. On October 14 the county hired engineers to review it. Here is what the filing says, what the site has been through, and what has to happen before the first slab is poured.

October 14, 2026 · 9 min read

AI Infrastructure

Summer of the Gigawatt: Four Campuses, Twelve Gigawatts, and the New Shape of an AI Data Center

In eight weeks this summer, OpenAI, Google, Amazon, and Brookfield surfaced four US campuses of a gigawatt or more in Georgia, Wyoming, Texas, and Kentucky. Four companies, one template: a code name, a single tenant, power phased in from 2028, closed-loop cooling, and increasingly a gas plant of their own.

September 3, 2026 · 9 min read

AI Infrastructure

Stargate on the Lakefront: Wisconsin's Biggest Power User Rises From Port Washington Farmland

Vantage Data Centers is building a $15 billion, 902 MW campus for OpenAI and Oracle in a small Lake Michigan port town. The buildings are going up fast. The power lines, the politics, and the town itself are having a harder time keeping pace.

August 31, 2026 · 6 min read

AI Infrastructure

Beyond NVIDIA: The Custom Silicon Quietly Filling AI Data Centers

Google, Amazon, Meta, Microsoft, and OpenAI are pouring billions into their own AI accelerators. Custom chips now claim more than a quarter of AI server shipments, even as NVIDIA posts record revenue.

August 22, 2026 · 5 min read

AI Infrastructure

The GPU-for-Rent Economy: Inside the Neocloud Gold Rush

Specialist GPU clouds like CoreWeave, Nebius, and Nscale are growing 200 percent a year on hyperscaler mega-deals and chip-backed debt. The open question is what the collateral will be worth when the loans come due.

August 20, 2026 · 5 min read

AI Infrastructure

One Computer the Size of a Building: How a Frontier AI Training Cluster Actually Works

Frontier AI clusters lash hundreds of thousands of GPUs into a single synchronous machine. Inside the NVLink domains, network fabric wars, oceans of fiber, and relentless hardware failures that define training at 100,000-GPU scale.

August 18, 2026 · 4 min read