BAKI,TurkicWorld
Nvidia and Broadcom are relatively insulated from the worsening power constraints facing US data centers, but delays in artificial intelligence deployments could put pressure on suppliers of memory, optical and other secondary semiconductor components, Morgan Stanley said.
The brokerage estimated that US data-center developers face a net power shortfall of about 34% through 2028, equivalent to roughly 32 gigawatts, even after accounting for measures such as on-site power generation and fuel cells.
The rapid expansion of generative AI has triggered billions of dollars in investment in data centers, but the enormous computing and electricity requirements of AI systems are increasingly colliding with constraints in the US power supply.
Morgan Stanley and Goldman Sachs have both highlighted growing challenges facing the US data-center construction boom. These include limited electricity availability, labour constraints and political opposition.
Morgan Stanley said it does not currently expect the power bottlenecks to put Nvidia’s or Broadcom’s 2027 forecasts at risk. The brokerage cited the companies’ visibility into chip deployment, geographic expansion and coordination between data-center operators, semiconductor suppliers and the electricity supply chain.
However, if customers are unable to deploy purchased chip capacity, they could delay deliveries or cancel orders, Morgan Stanley warned.
Memory, optical, power-management and analog semiconductor suppliers could be among the most exposed to inventory disruptions if AI infrastructure projects are postponed.
The developments highlight a growing challenge for the AI industry: demand for computing capacity is rising rapidly, while the physical infrastructure required to power increasingly energy-intensive data centers is struggling to keep pace.
