How Energy Investors Are Riding the AI Wave? Betting on Power Stocks!

With continuous advancements in AI, particularly in the efficiency of hardware like Graphic Processing Units (GPUs), the cost of running GenAI algorithms in data centers is significantly decreasing. The primary drivers of this cost reduction include: 1) Enhanced hardware efficiency; 2) Optimized data center design; 3) Economies of scale; 4) Technological progress. For instance, compared to the Hopper GPU, the Blackwell GPU is expected to reduce the capital cost per teraFLOPs for data centers by about 50%.

The decline in computing costs is set to fuel a wider adoption of GenAI technology across various domains, including data centers, cloud computing services, and edge computing, as well as spurring the expansion of data center infrastructure.

According to Morgan Stanley's updated forecast on global data center power demand, which includes the varying amounts needed for different types of power generation, the baseline scenario projects that by 2024 and 2027, global data centerpower demand will be approximately 430 and 748 terawatt-hours (TWh), respectively. These figures correspond to about 2% and 4% of the total global electricity demand in 2022.

Meanwhile, the compound annual growth rate (CAGR) for power demand fromGenAI is expected to be around 105% from 2023 to 2027. During the same period, the CAGR for global data center power demand, including that for generative AI, is estimated at about 20%. As a result, the growth of power infrastructure is falling far behind the surge in demand for electricity from GenAI.

The challenges facing the growth of power infrastructure include insufficient capacity of existing transmission lines, project delays due to complex planning and permitting procedures, supply chain bottlenecks, the need for new technologies and innovation, substantial capital investments, changing policy and regulatory environments, public acceptance issues, as well as geographical and climatic constraints.

Consequently, Morgan Stanley suggests that energy investors should focus on companies capable of providing reliable and efficient power solutions, as these entities are likely to benefit from the rapid expansion of data centers. The firm has increased the target prices for a host of power companies, identifying the following targets: