AI is fueling global power demand. Morgan Stanley sees opportunities for these 'underappreciated' winners

Published Thu, Feb 1 202411:06 AM ESTUpdated Thu, Feb 1 202411:13 AM EST

Generative artificial intelligence is about to rapidly scale up global power demand. That's a boon for several power providers and data center builders. Morgan Stanley highlighted more than a dozen overweight-rated stocks that stand to benefit from the surge in generative AI power demand, which is increasingly being driven by the need to build and supply data centers that train and deploy advancing AI models . Power demand from generative AI will soar from 46 terawatt hours to 224 TWH between this year and 2027, according to the firm's base case scenario. That's equivalent to more than 75% of the total global data center power use in 2022, or close to Spain's entire power consumption in 2022, the firm said. "We expect growth upside for power providers and data center infrastructure stocks, but do not expect GenAI power to move the needle for regulated utilities," Morgan Stanley analysts wrote in the Monday research note. "We believe the rapid power demand growth from GenAI is not well understood, and not priced into a number of stocks." To be clear, many of Morgan Stanley's picks are large-scale renewable energy and infrastructure stocks that stand to gain market share from this trend over time. CNBC recently reported on the companies that already specialize in servicing data centers' physical infrastructures and the power grids that supply these power-hungry sites. Here are some of the firm's top picks: Sustainable data centers Morgan Stanley's picks favor companies such as Bloom Energy , AES Corp and NextEra Energy , that provide scalable renewable energy because this will be an important focus for new data centers. That's because tech giants are trying to reach sustainability goals at the same time their businesses will need more energy to fuel generative AI. Google, for example, plans to use 100% carbon-free energy in its data centers by 2030, while Microsoft is aiming to operate on entirely renewable energy sources by 2025. Morgan Stanley's base case is that, if 80% of incremental data center power demand in the U.S. is from renewable energy, developers would need to build about 5 gigawatts of renewables this year, which should be manageable in most cases. "We think the impact on global carbon emissions is likely to be small, while the sustainability benefits of GenAI are likely to be large," the analysts said. Bloom Energy has "exceptionally high upside to (as yet unrealized) orders from data centers," the Morgan Stanley analysts wrote in the note. That's because the company provides baseload power fuel cells — devices that generate electricity and heat by combining fuel with oxygen — to data center customers. "BE is a secular winner as the world adopts GenAI given the significant incremental power demand associated with data center growth, coupled with grid infrastructure constraints and reliability concerns," analyst Andrew Percoco wrote in a Wednesday note where he called the stock an "underappreciated AI winner." Percoco estimates that generative AI could require 20 gigawatts of global electricity generation capacity through 2027. He expects Bloom could generate nearly $75 billion in near-term revenue for its fuel cell product. Bloom shares could rise more than 94% from its latest close, according to his price target. However, not all analysts agree, and the stock has lost about 18% so far this year. Bank of America recently downgraded shares to underperform from neutral, saying revenue will likely be about flat from 2023 to 2025 due to low orders. Among large-scale renewable developers in the U.S., AES has the largest commercial and industrial market share, according to Morgan Stanley. Percoco thinks AES' commercial relationships with Big Tech, including Google, Microsoft and Amazon, could be a strength in serving data center demand as these tech giants are considered to be the data center "hyperscalers." He estimates, however, that generative AI currently accounts for 0% of revenue, but that it could grow to 3.5% by 2027. AES shares have plunged roughly 13% since the start of the year amid a larger slide in utilities stocks. Morgan Stanley sees an upside of nearly 56% over the next 12 months, however. NextEra , the leading renewable developer in the U.S., is "well-positioned" to supply renewable power to new data centers, according to Morgan Stanley. The firm expects NextEra could go from incurring 0% of generative AI-driven revenue to 2.2% by 2027, especially given the company's geographic scale, competitive costs with its purchasing power, and its ability to integrate wind, solar, storage and software to optimize its services. Last week, NextEra reaffirmed its full-year guidance and posted a beat on quarterly earnings per share, excluding items. Morgan Stanley analysts think the stock can gain more than 30% from its latest closing price. Stocks supplying data center infrastructure Morgan Stanley also sees an opportunity for vendors that provide power management solutions like Delta Electronics and cable manufacturer Prysmian , which analyst Max Yates called a global grid play. "As power demand grows, in order to reduce costs, we think customers will turn to IT equipment such as Delta's to support power efficiency and for reliable backup power," analyst Sharon Shih said. She pointed to Delta's InfraSuite software product, which optimizes critical device monitoring in data centers. Other stocks that stand to benefit from data center infrastructure needs of data center are real estate investment trust company Prologis , which is set to build out between $7 billion and $8 billion worth of data centers over the next five years, and Mitsubishi Electric , which builds a variety of products and systems for data centers related to air conditioning, energy savings and other functions. Apart from data center infrastructure and power providers, utilities are another part of the data center servicing ecosystem. But it's the only business model that is not reliant on power demand for its growth, Morgan Stanley analysts said, as the power needed for generative AI is largely relative to the size of power providers, which are often small companies. Analysts noted that regulated utility companies, such as Dominion Energy , provide power infrastructure at a "massive scale, and data center power needs are still going to be a small part of utility capex budgets."