The objective of the proposed research is to mitigate the technical challenges of frequency instability and network congestion introduced by the rapid expansion of energy-intensive data centers. The work addresses these issues from two different perspectives. At the device level, it presents a novel control design for grid-forming resources using an adaptive internal model principle. This scheme utilizes real-time frequency estimates to ensure zero steady-state tracking error, maintaining stable inverter operation even during abrupt frequency deviations caused by the tripping and reconnection of large-scale data centers. At the network level, it enhances real-time congestion management through network topology optimization. By utilizing a linear AC power flow model rather than the standard DC approximation, the research provides a more accurate framework for transmission switching decisions in real-time operations timescales. Preliminary results from a network topology optimization application (wildfire risk minimization) suggest that this model can lead to more realistic switching decisions, considering the actual network physics more accurately.