After months of no progress, several long-paused Microsoft self-build campuses are finally showing signs of life. Conover, North Carolina broke ground in Q1 2026 after an extended pause. Chase City, Virginia showed what appears to be construction activity in the same quarter. Castroville, Texas is also showing early signs of movement after more than a year with little visible progress. Pictured: Microsoft’s campus in Conover, North Carolina (April 2026) (1/3)🧵

Datacenter developers are increasingly planning projects in unincorporated county land, and it's not an accident. Outside city limits, they can sidestep city council approvals, municipal zoning votes, and urban land-use reviews. This is redrawing the map of where large-scale AI infrastructure gets built. (1/4) 🧵

AI-focused datacenters are moving beyond the traditional “plug into the grid” model. At multi-hundred-MW and gigawatt scale, they’re starting to look like energy projects first. ⚆ In West Texas, Google is executing a multi-site expansion tied directly to power strategy. South Haskell County alone pairs a ~440 MW datacenter build with 640 MW of solar, while other Google-linked campuses layer in wind, storage, and behind-the-meter generation. ⚆ In Pennsylvania, the former Homer City coal plant is being redeveloped into a 4.5 GW gas-powered datacenter campus, with H-class turbines permitted and construction underway. ⚆ In Tennessee, xAI is deploying turbines directly alongside its Colossus datacenters to support rapid ramp while permanent power infrastructure is built out. Historically, datacenters were about land, fiber, and tax incentives. Today, power availability is the site strategy. Cooling, turbines, substations, fuel logistics, and permitting now decide where AI gets built – and how fast. Datacenters aren’t just customers of the grid anymore, they’re becoming energy infrastructure in their own right.

Why Abu Dhabi for Stargate UAE? When you're building a 5 GW AI datacenter campus, the choice of location comes down to energy capacity, infrastructure readiness, and room to scale. While the final site remains undisclosed, two potential locations stand out. Here's a breakdown of why.

Is OpenAI’s ‘Breakup’ with Microsoft Redefining the AI Infrastructure Landscape? Behind Microsoft’s datacenter slowdown is something deeper than just market recalibration. It’s about losing control of the very AI workloads it helped scale. Below we explain why. 🧵👇 1/4