In the news. How much of the AI infrastructure boom reflects real demand? Reuters reports that power requests—driven largely by data centers—in some US regions exceed ten times the industry’s estimate of current data center power use. But some projected demand fell away when utilities introduced deposits and other tests of project commitment. As “ghost” demand complicates generation and transmission planning, regulators are tightening disclosure and permitting rules. For infrastructure leaders and investors, credible project pipelines will be critical to deciding where to build—and where to put capital. [Reuters]

Investment will pay off only if demand for AI services grows fast enough to fill the computing capacity being built for it—the chips, power, and data centers.

On McKinsey.com. AI’s bottlenecks shift as technology, infrastructure, and organizations advance at different speeds, explain Arvind Govindarajan, María Jesús Ramírez, Shubham Singhal, Tanguy Catlin, and Xavier Rigby. With nearly $7 trillion in estimated global data center investment needed through 2030, leaders will have to understand which constraints could shape the next phase of the build-out. Power and grid connections are major bottlenecks today, but applications, workforce skills, and organizational workflows could be next—requiring leaders to track these shifting constraints and redirect investment as conditions change, rather than extrapolate growth from AI capabilities alone.

Spot AI’s next bottleneck