Power Constraints Pose Risks to AI Chip Supply Chain Despite Sparing Major Suppliers

Morgan Stanley projects that US data centers face a 34-gigawatt power shortfall through 2028, yet predicts that leading chip manufacturers Nvidia and Broadcom will have limited exposure to these electricity constraints due to their visibility into deployment plans and coordination with infrastructure providers. Secondary semiconductor suppliers manufacturing memory chips, optical components, and analog products are considered more vulnerable to potential delays if data-center projects are postponed due to power limitations. The analysis underscores emerging disparities within the artificial intelligence hardware market, where industry giants may weather infrastructure bottlenecks while smaller component makers face greater risk.

The semiconductor industry faces an emerging infrastructure challenge as artificial intelligence expansion strains US electrical capacity. Research indicates that data center operators will encounter a substantial electricity deficit over the coming years, creating potential delays across hardware manufacturing and deployment timelines. This bottleneck threatens to disrupt the supply chain unevenly, with consequences that vary significantly based on company size and market position.

Smaller component manufacturers specializing in memory, optical systems, and power management face disproportionate vulnerability to project delays. Unlike major chip designers with direct relationships and advance notice from major clients, these secondary suppliers lack visibility into deployment schedules and infrastructure coordination efforts. Consequently, demand fluctuations resulting from postponed data center projects could leave them managing unexpected inventory levels and revenue disruptions.