A report by Goldman Sachs reveals that the critical bottleneck hindering AI development is not capital but rather the availability of power supply. Data indicates a swift surge in global electricity demand for data centers, with a projected 50% increase in total consumption by 2027. Significantly, 60% of this demand will necessitate the construction of new power generation capacity. By 2030, the demand is expected to escalate further, reaching a staggering 160% of current levels. Tyler Miller, Head of Global Power and Utilities at Goldman Sachs, emphasized the need to focus on nuclear power, noting that substantial investments have already been channeled into the commercialization of small modular reactors and nuclear fusion technology. Nevertheless, given their lengthy construction periods and substantial costs, there remains a necessity for tailored capital solutions or governmental support to mitigate associated risks. Rebecca Kruger, a Partner in Goldman Sachs' Natural Resources Department, remarked that the era of stable power demand has concluded, with data centers now serving as the principal driver of industry growth. Consequently, a substantial upswing in large-scale infrastructure investment is currently underway.