Our Favorite Long-Term Themes: Power Demand & Grid Infrastructure
By Clear Perspective Advisors on August 14, 2024
A Weekly Wisdom Series
This week’s Weekly Wisdom will be the first installation of a multi-week series in which we will highlight our favorite long-term investment themes. With the recent market volatility, we find it pertinent to go back to the basics and focus on what we know. The themes we will be discussing over the next number of weeks are ones we have deep conviction in – ideas that will be having material impacts on our economy over the next five to ten years. We will discuss why we find value in these themes and how we are positioned to gain from them.
Power Demand & Grid Infrastructure
As one of the top themes we are positioned for, it is a no-brainer that a mass increase of investment into energy infrastructure will be needed to complement the jumps in electricity and energy demand from recent technological advancements. Mega-cap technology companies are in an arms race to be the first to take advantage of these emerging technologies through billions of dollars of investments in artificial intelligence (AI), generative artificial intelligence, cloud and data center infrastructure, and research and development towards new innovations.
Amazon (AMZN), Alphabet (GOOGL), Meta (META), and Microsoft (MSFT) are allocating massive amounts of capital to these technologies – nearly $200 billion this year with a combined 41% y/y increase in capital expenditures. For context, in aggregate, the S&P 500 is expected to grow capex 9% y/y in 2024.
Chart 1: Amazon, Alphabet, Meta, and Microsoft Expect to see a 41% Jump in Capital Expenditures in 2024[1]
Mega-cap tech companies are making sure they are ahead of this new technological cycle, and would rather spend too much than too little in developing AI. Alphabet’s CEO, Sundar Pichai, stated in the company’s most recent earnings call that “the risk of under-investing is dramatically greater than the risk of over-investing”. Investments in data centers to house these technologies are some of the largest investments being made, as GOOGL has said that 50% of its capital expenditures is going towards data centers with the other half focused on generative AI.
Booming Growth in Power Demand
The need for greater energy sources will skyrocket following the full rollout of these technologies. AI takes an immense amount of energy to operate. For example, ChatGPT queries are 6-10x more power intensive than Google searches, and OpenAI spends nearly 12 cents per 1,000 words ChatGPT generates due to computing costs.[2]
According to Goldman Sachs, U.S. data center power demand is expected to grow from 146 gigawatts today to 397 gigawatts by 2030. Globally, data centers consumed an estimated 460 tera-watt hours of electricity in 2022, and it is expected that by 2026 this number could reach more than 1,000 terra-watt hours – roughly equivalent to the total electricity consumption of Japan.[3] To put this in perspective, one gigawatt is one billion watts, and one terra-watt is one trillion watts. One gigawatt is equivalent to the energy output of 310 wind turbines or 100 million LED bulbs.[4]
Wells Fargo projects AI-specific power demand to surge by 550% over the next two years, and realize 8,050% growth by 2030, from a total usage of 8 terra-watt hours in 2024 to 652 terawatt hours by 2030. NVDA’s Blackwell, the company’s newest GPU, uses nearly 1,200 watts of energy consumption – a 300% increase in power consumption from the previous model. [5] These chips boast better performance-per-watt, but use more energy in aggregate.
As the economy leans more heavily on AI chatbots, and companies such as AAPL fully integrate AI technologies into their products, energy demand, and broader data center usage will grow. This will also provide a network effect to other sectors such as materials, utilities, and industrials as new construction projects are started, energy usage increases, and materials like copper and nickel begin to see greater demand to build necessary infrastructure.
Chart 2: Data Center Power Demand is Likely to Triple by 2030[6]
In total, tech companies are heavily focused on investing in this new technology cycle. Energy demand is likely to skyrocket, and infrastructure will be needed to house the hardware. The U.S. power grid is also in need of a full revamp to assist on the supply side, as the North American power grid is currently over 40 years old. Power demand and grid enhancement is a multi-year, and possibly decade, theme that we are positioned for across our portfolios.
How We Are Exposed to This Theme
Many sectors of the economy will be affected by the tailwinds from increased power demand, data center growth, and an improved energy grid. Mega-cap tech will continue to invest heavily in innovative technologies relating to AI, which will be met with increased power needs. Quanta Services (PWR), Eaton (ETN), and GE Vernova (GEV) are companies with exposure to an expanding grid and power demand. PWR and ETN have immense backlogs of mega projects (infrastructure projects >$1 billion) and will continue to benefit from the need for more data centers and large-scale energy solutions. GEV is one of the world’s leading energy suppliers, which will benefit from a greater demand for energy. GEV’s installed base of gas and wind turbines as well as other power sources generates ~30% of the world’s total electricity.
Craig Arnold, CEO of ETN, mentioned on its most recent earnings call that over 90% of data center orders they are receiving are “conventional” data centers, not tied to AI. He said the global backlog for data centers is close to $140 billion, equating to eight years of production, and that the AI inflection point still has “yet to be hit” across data center projects. ETN projects a 25% annual growth rate in data-center-related demand through 2025.
Chart 3: Orders for Eaton’s Data Center Equipment Has Doubled in the Past Year[7]
Across the utilities sector, Dominion Energy (D) and Black Hills (BKH) offer diversification benefits from other sectors while maintaining exposure to data centers and power demand. D connected nine new data centers YTD through July and expects to connect 15 in total for 2024. D also experienced six new all-time high peak energy demand records in July, attributable to economic growth, electrification, and data center expansion. July’s records beat previous demand records set in December 2022. BKH recently partnered with Meta to power its new AI-focused data center in Cheyenne, Wyoming, which is expected to begin service in 2026. BKH has extensive plans across many states to expand electrical systems and enable solutions across data center and blockchain customers.
On the hardware side, Broadcom (AVGO) and Seagate Technologies (STX) benefit from tailwinds in the technology space, mainly semiconductor chips, AI data centers, and the need for more cloud and compute storage. AVGO saw its AI revenues grow 280% y/y in the previous quarter driven by strong demand from AI networking and custom ASICs they provide GOOGL, META, and ByteDance. AVGO is also seeing immense strength in VMware, a software business it recently acquired that produced $2.1 billion in revenue in Q1, $2.7 billion in Q2, and is accelerating to a $4 billion revenue run rate per quarter.
STX is a prime beneficiary of the need for more data storage, which will continue to grow with AI spending. STX’s leading position in the hard disk drive market has allowed it to flex pricing power, where its gross margins grew by 1,140 basis points y/y in the previous quarter. STX will continue to benefit in a supply-constrained environment with demand moving higher as cloud customers are expanding its capacity.
Lam Research (LRCX) is another semiconductor chip beneficiary of the current environment. The company recently grew its systems revenue by 27% driven by the strength in the memory market. LRCX and its peers have noticed an uptick in Fab utilization rates which is front-running demand for generative AI products. Lam has positioned itself to benefit from higher leading-edge spending as well as advanced packaging.
International Business Machines (IBM) is unique because it benefits in multiple ways as it is exposed to hybrid cloud/software, consulting, and infrastructure spending. IBM’s ability to capture accretive M&A is continuing to benefit its software packages. WatsonX grew its book of business by over $1 billion sequentially as IBM is capturing demand for gen AI, they have now grown their gen AI book of business by just under $2 billion in the past 3 quarters.
There are several ways to position the increase in power demand and infrastructure across the technology, industrial, energy, material, and utility sectors. We focus on industry leaders with strong balance sheets and management teams that possess attributions to continue growing their top and bottom lines at a solid pace. We believe power demand will continue to expand over the coming years to support the innovative technologies that are entering the market.
[1] Source: FactSet. As of August 12, 2024.
[2] Source: New York Times. As of May 10, 2024.
[3] Source: IEA. As of January 2024.
[4] Source: Energy.gov. As of August 24, 2023.
[5] Source: Forbes. As of June 20, 2024.
[6] Source: Goldman Sachs. As of May 14, 2024.
