Data Center Knowledge

As AI and high-performance computing (HPC) drive up server power densities, data centers are turning to liquid cooling to remove heat more efficiently. Rack densities have jumped from 5-10 kilowatts (kW) to 30 kW, 50 kW, or even higher – levels where air cooling struggles. More power means more heat, and liquid cooling captures it far more effectively at the source.

The Department of Energy (DOE) has ordered PJM Interconnection and Constellation Energy to keep 760 MW of generation at Pennsylvania’s Eddystone Generating Station available to operate through Nov. 20, extending an emergency measure as the grid faces rising electricity demand and a wave of generator retirements. The DOE’s reliability analysis includes roughly 25 GW of projected PJM load growth, including 15 GW from data centers, against about 17 GW of announced fossil-fuel retirements. The Aug.

Lancium is partnering with Nvidia to deploy the chipmaker’s AI factory technology across a portfolio that the Texas infrastructure developer says includes 4 GW of leased capacity and more than 15 GW of powered land in development. The partnership anchors Nvidia’s role in designing and deploying AI infrastructure across Lancium’s portfolio.

Data center construction remains the defining infrastructure story of 2026. In a broader construction market showing signs of softness, digital infrastructure is a robust exception. Citing the Dodge Momentum Index, Construction Dive reported that data center construction planning momentum picked up in July, “again shining as the brightest sector in building.” This pipeline reflects demand for insurance and risk management expertise in technically complex, capital-intensive projects.

Prometheus Hyperscale is betting that the fastest way to put AI compute in West Texas is to avoid an initial ERCOT grid connection altogether. The company plans a 1.5 GW data center campus in Reeves County, Texas, powered by an islanded generation system. The project has a development path to 2.5 GW, with pipeline natural gas supplying the initial generation and a planned co-located fractionation facility providing ethane fuel flexibility after it enters service.

Power availability is becoming a strategic development factor Power availability has become the primary limiting factor for new U.S. data center development. AI-driven load growth is accelerating faster than grid expansion, resulting in congestion, long interconnection queues, and increased project risk.[i] As a result, power has evolved from a development constraint into a strategic consideration.

Telecommunications firm Zayo is reserving manufacturing capacity with Corning for a material portion of the fiber it expects to need through 2030, protecting plans to build 15,000 route miles of new and overbuilt infrastructure as AI drives demand for data center connectivity. The agreement covers part of the telco’s broader funded network expansion, which includes 28 new or in progress routes, 10 new routes, and 18 overbuilds.

EdgeCore Digital Infrastructure says it will fund the generation, transmission, and delivery infrastructure its expanding data center campuses need. At the gigawatt scale, that exposure can reach hundreds of millions of dollars, but the harder question is where utilities and regulators draw the line between specific data center costs and broader grid upgrades.

PJM Interconnection envisions a power system in which data centers drive most new electricity demand, but its five-year strategy begs a critical question: how much responsibility should the regional grid operator assume for turning uncertain load forecasts into generation, transmission and reliability commitments? PJM’s Board of Managers approved the strategy in July, and PJM released it Aug. 19.

Worldwide data center capital spending is projected to surpass $3 trillion by 2030 as hyperscalers, sovereign AI programs, and specialized cloud providers expand infrastructure to support growing AI workloads, according to a new report from Dell’Oro Group. The research firm said its 2030 capex outlook has nearly doubled since its January 2026 forecast, reflecting higher hyperscaler spending guidance, increased estimates for global data center power capacity, and rising commodity costs.