This Week in Data Center Infrastructure — May 12, 2026

A data center developer is planning a new facility in Broken Arrow, Oklahoma, with city officials committing to transparency throughout the approval process. The project signals continued regional expansion for hyperscale and colocation infrastructure in the South Central US.

Early-stage site selection in growing markets means crews need to prepare for fast-track mobilization once permitting closes—staging material yards, coordinating utility cuts, and locking cabling pathways before ground breaks.

A Michigan utility has halted new power connections to data center projects for one year, effectively freezing two potential facilities in the region. The moratorium reflects growing grid capacity constraints in markets with heavy AI infrastructure demand.

Projects already approved but not yet energized face timeline uncertainty—crews must lock in cable runs, rack staging, and commissioning schedules with utilities now, or risk extended delays in infrastructure deployment and final closeout.

Core Scientific is building a 1.5 GW AI campus in Oklahoma and expanding with another facility in Texas, repositioning crypto-era data center assets for high-density colocation and AI workloads. The operator aims to deliver roughly 3 GW of AI and colocation capacity across both sites.

Retrofit and repurposing of existing facilities requires detailed cable audits, infrastructure refresh planning, and coordination between decommissioning old systems and installing new fiber, power distribution, and cooling pathways—plan staged crew rotations and material sequencing carefully.

AMD reported strong revenue growth driven by EPYC and Instinct chip demand for AI infrastructure, signaling sustained investment in hyperscale deployments. Inference workloads are expanding hardware refresh cycles and driving new rack-and-stack activity across multiple regions.

Rising inference GPU deployments mean higher cable density, denser rack configurations, and more frequent hardware swaps—ensure your cabling design supports high-count GPU cluster layouts, plan for increased smart hands labor, and schedule hardware refresh crews with buffer capacity.

Data center work is accelerating across every market — new builds in secondary cities, retrofits of crypto-era facilities, and GPU-dense AI deployments all landing at once. The execution risk is real: compressed timelines, fast-track mobilization, and infrastructure refreshes that require careful sequencing. AES is built for exactly this environment — trained crews, clean documentation, and reliable closeout.

Need field execution support? AES supports data center and IT infrastructure projects across the US and Canada — structured cabling, fiber, rack-and-stack, smart hands, hardware refresh, and decommissioning. Reply to this email or visit apexsolutions.io to start a conversation.