A report from Morgan Stanley reveals that the ongoing construction of AI infrastructure across the United States is propelling power demand into uncharted territory, with the potential for power supply capacity to emerge as a significant bottleneck for AI industry growth. Projections indicate that by 2028, the cumulative power consumption of data centers in the U.S. will soar to 69 gigawatts, resulting in a power deficit of roughly 44 gigawatts. To mitigate the strain on power supplies, the Department of Energy is gearing up to extend financial support to nuclear power initiatives. The scarcity of power resources could hinder the execution and pace of AI investments, potentially prolonging construction timelines. Morgan Stanley has put forth the 'Time to Power' strategy, which, upon full implementation, is anticipated to curtail the power shortfall to about 13 gigawatts by 2028. This strategy encompasses natural gas turbine ventures, Bloom Energy's fuel cell technology, direct power purchase agreements with existing nuclear power facilities, and the repurposing of Bitcoin mining operations. Among these, the conversion of Bitcoin mining sites and Bloom Energy's fuel cell systems present notable benefits. The report also foresees the rise of varied power trading frameworks and scrutinizes two business models for transitioning Bitcoin mining facilities into high-performance computing (HPC) data centers, recognizing both as having the potential to generate value. Moreover, it supplies valuation benchmarks for the conversion of mining sites. In conclusion, the report issues a word of caution regarding market risks and advises investors to exercise prudence.