Trump Moves to Make Tech Giants Pay for Surging Power Costs

Summary

Mandeep Singh of Bloomberg Intelligence discusses the massive power requirements of AI data centers, noting that new facilities could need 10 times more electricity than existing ones and that an aging grid may struggle to supply it. He warns that AI workloads are structurally lower-margin for hyperscalers like Amazon, Microsoft, and Google, which will pressure their margins. He also says AI demand is creating shortages in previously neglected semiconductor components like storage and CPUs, with Intel benefiting from the tight supply environment.

• AI data centers may require up to 10x more power than existing facilities.
• The 70-80 year old U.S. grid faces challenges supplying that power.
• Higher electricity prices are already reflecting the realization of AI power demand.
• AI workloads carry sub-50% gross margins, below traditional public cloud and on-premise software.
• Hyperscalers Amazon, Microsoft, and Google may see margins compress.
• AI demand is causing shortages in storage, CPUs, and other components.
• Intel is cited as a beneficiary of the tight AI semiconductor supply environment.
• IT infrastructure refresh cycles could add to component shortages.