An investment bank report issued Monday stated that while Nvidia and Broadcom are positioned to withstand near-term power availability issues in U.S. data centers, the broader supply chain for artificial intelligence components faces potential disruption. The Morgan Stanley analysis concluded that any slowdown in AI infrastructure deployment caused by power constraints could negatively impact manufacturers of memory, optical components, and other related semiconductor products.
The report addresses the growing energy requirements of data centers designed for intensive AI workloads. According to the analysis, the demand for electricity to power advanced processors, such as GPUs, is creating a bottleneck for new data center construction and expansion in several U.S. markets.
Morgan Stanley’s assessment identifies Nvidia and Broadcom as “relatively insulated” from these immediate pressures. The bank did not detail the specific reasons for this insulation in the note. The focus of the report was on the secondary effects of the power crunch.
According to the analysis, if companies are forced to delay the launch of new AI data centers due to an inability to secure adequate power, the demand for a wide range of components would be deferred. This includes high-bandwidth memory (HBM), optical transceivers for high-speed networking, and various other chips that populate server infrastructure. The report suggests that this potential for delayed deployments represents a risk for companies that supply these secondary components.
The findings from Morgan Stanley contribute to an ongoing discussion among technology industry leaders and utility providers about the infrastructure needed to support the expansion of artificial intelligence. Public statements from energy and technology companies have previously highlighted the significant increase in electricity consumption driven by the mass adoption of AI technologies. The report frames the power supply not as a technological barrier but as a logistical one that could alter the timeline for AI hardware rollouts.
