U.S. data centers could face a 33-gigawatt power shortfall through 2028, according to Morgan Stanley’s latest estimate, even after accounting for accelerated solutions such as onsite generation, fuel cells and nuclear power.

The bank raised its estimate for incremental U.S. data-center demand from 68 GW to 97 GW. Its model counts 21 GW of facilities already under construction and 19 GW of additional grid capacity, plus probability-weighted alternative power sources.

The shortage is being amplified by increasingly dense AI hardware. Morgan Stanley models an Nvidia Vera Rubin rack at roughly 234 kW and Rubin Ultra at approximately 600 kW per rack, including memory, networking, power delivery and liquid cooling.

That means better computing efficiency does not necessarily translate into lower total electricity consumption. Morgan Stanley expects tokens per watt to improve nearly sixfold from 2025 through 2028, while rapidly expanding AI deployment pushes absolute power demand higher.

The forecast connects directly with BitcoinVersus.tech’s recent coverage of Alibaba’s 20 GW data-center target. Increasingly, the AI infrastructure race is measured not only in GPUs, but in megawatts, gigawatts, substations and time-to-power.

Source: Yahoo Finance.

BitcoinVersus.Tech Editor’s Note:
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. If you would like to support to help further secure the integrity of our research initiatives, please donate here: 3C9o19EH5HSiwEPyCTmEKzxhNCbo2X6TTb

BitcoinVersus.tech is not a financial advisor. Information is provided for informational purposes.