The buildings that house artificial intelligence are being rethought from the ground up, and the reason is electricity.
According to Bloomberg, AI players are reimagining the data center from the ground up as soaring energy demand already strains existing infrastructure. The computing power behind modern AI consumes vast amounts of electricity, and the facilities originally designed for ordinary web and cloud workloads were never built to handle that kind of load. Rather than simply adding more machines to old designs, the industry is reconsidering the fundamentals of how these centers are laid out, cooled, and powered.
One sign of that shift comes from the equipment side. According to Yahoo Finance, ABB is expanding its partnership with Nvidia to boost AI data centers. ABB is an industrial technology company known for electrification and power equipment, while Nvidia makes the chips that have become the engine of the AI boom. A deeper tie between a power-and-electrification specialist and the leading AI chipmaker points to where the bottleneck now sits: not just in the processors themselves, but in delivering enough reliable power to keep them running.
Together, the two developments sketch a single story. As AI workloads multiply, the constraint is shifting from raw computing capacity toward the physical systems that feed and cool that computing — the wiring, the grid connections, and the heat management that determine whether a facility can actually operate at scale.
Why it matters: the pace of AI progress increasingly depends not only on better chips, but on whether the world can build and power the data centers to run them.