October 2024
At the beginning of 2023, Google Trends showed a sharp spike in Americans searching for the term “artificial intelligence.” Around the same time, the stock price of AI poster child Nvidia began to quickly rise. In February 2023, the cover story on Time Magazine was entitled, “The AI Arms Race is Changing Everything.” By last summer, we began to hear from our portfolio companies about how AI was impacting their business models, as well as from clients wanting to know if they had exposure to this burgeoning theme. Investors frantically bid up the stock of almost any company that was perceived to benefit from this new technology. Just like that, a wave of AI frenzy had washed over the markets.
Much like with surfing, catching an investment wave isn’t the easiest thing to do. You can be too early and wait in cold water for a long time, often losing patience and heading back ashore before the wave comes in. One can also be too late and miss most of the ride, or worse yet, wipe out. It is also possible to grab the wrong wave – one that breaks too early or never fully forms.
As any experienced surfer knows, sometimes a good wave is followed by another good one… and occasionally it is better than the first. Though we cannot be sure, there appears to be a large one forming out on the horizon for astute investors. The theme of this wave is electrification, and we would speculate that you’ll be hearing a lot more about it in the coming quarters.
Electrification is the concept that electricity’s role as a source of power is set to expand across a variety of sectors. Growth in U.S. electricity consumption has been stagnant over the past 20 years. In the year 2003, the U.S. consumed 3.66 billion kilowatt hours (kwh) of electricity.1 By 2023, U.S. energy consumption had increased to only 4.10 billion kwh, representing a compound annual growth rate of 0.6%. Power industry experts cite a number of reasons for this slow growth, including technological advancements that have made everything from light bulbs to HVAC systems more energy efficient; growth in renewable energy sources, such as solar and wind; and changes in consumer behavior, such as remote working.
Tide together
Yet, the artificial intelligence wave is helping to form the second wave in our analogy – that of electrification. According to the International Energy Agency, a single Google search requires 0.3 watt hours of electricity, as compared to a single ChatGPT query (an AI-powered search), which uses 2.9 watt hours of electricity – a near tenfold increase in energy consumption.2 While you may not be consciously using an AI product like ChatGPT right now, rest assured, you’ve already been using AI. Whether it’s your iPhone completing sentences for you or Netflix making viewing suggestions, AI is working behind the scenes and quickly spreading in its applications.
To utilize AI, companies must buy new equipment in the form of computers, servers and switches. This hardware is stored in a data center, a large building that is filled with rows of IT equipment and is properly secured and cooled to make certain that the technology runs smoothly all the time. Much like aggregate U.S. electricity consumption, energy usage by data centers hasn’t changed much for the past decade, as greater usage of technology by our population has been offset by efficiency gains in computing. That trend though too, appears set to change with the emergence of AI. Currently, data centers account for ~3% of U.S. electricity consumption. Goldman Sachs estimates that by 2030, data centers will account for 8% of our country’s electric demand, helping overall electricity demand to grow by approximately 2.5% annually over the next decade.
That growth may not sound overwhelming, but given the size of the U.S. electric grid, substantial investments will be needed to manage this expansion. I’m sure many of you have read articles about problems with our current infrastructure. Even at current demand, service is becoming less reliable due to a multitude of factors, including uneven population growth (particularly around Sun Belt cities), the environment (floods, wildfires and heat waves) and security breaches. The power drain from AI technology, as well as crypto mining (another process requiring significant computing power), will only exaggerate the weaknesses in our nation’s grid.
Texas may serve as a precursor to what many parts of the country may experience going forward. A decade ago, data centers utilized a negligible share of the state’s energy demand. This year, data centers are expected to consume 4% of the state’s electricity. By 2030 – just a little more than five years from now – Bloomberg NEF expects data center demand to comprise 16% of the state’s needs.3 The Electric Reliability Council of Texas (ERCOT) states that electrical demand is already dangerously close to the state’s total electric supply, even assuming no disruptions.
An epic swell?
Given the rapid development of AI technology, we expect a mad scramble by both the public and private sector to expand electric capacity across the country. Companies that supply equipment such as transformers, switching gear, power lines and circuit breakers should see strong demand for the foreseeable future. Likewise, the entire supply chain for providing electricity should benefit – that includes anything from sourcing natural gas (which is used to power many electric plants) to constructing additional power-generating facilities.
This hypothesis is not completely new to investors, as many of our holdings that have exposure to the electrification theme have witnessed strong gains over the past year. Similarly, the S&P 500 Utilities Sector is up 26% year to date, which would be the biggest annual gain for the sector in the past 20 years. While we are hesitant to stay on the wave too long, we do believe that this theme is far from over. Unlike many industries, growing capacity in the electricity market can take years. A power plant, for instance, typically takes five to seven years to construct.
In closing, we would encourage investors to keep an open mind when it comes to proper positioning on these waves of innovation. There’s no need to get too aggressive; likewise, one shouldn’t entirely avoid them. We remain confident that artificial intelligence is a game-changing technology that will reshape how we work and play for decades to come, much like the internet did decades ago. Electrification is a wave that we like as much as the AI one, if not more, simply because of its linkages to AI while carrying considerably lower valuations to participate.
We would caution against getting too excited about missing a “winner.” We’re still in the early chapters of the AI story and there will be plenty more to come. As evidence, consider AOL, the internet darling whose market capitalization peaked at $200 billion in 2000, only to be sold for $4 billion to Verizon in 2015. Likewise, recall the mobile phone revolution of the mid-2000’s. By 2007, Nokia’s stock chart had gone nearly straight up for several years as the company amassed a 40% global market share in mobile phones. Now, 14 years later, Nokia’s stock is down 81%. In short, this surfing business isn’t easy. Safety first and trust that the right wave will come.
All market data is sourced from Bloomberg unless otherwise noted.
1. http://large.stanford.edu/courses/2012/ph241/druzgalski2/docs/aer.pdf
2. https://www.goldmansachs.com/insights/articles/AI-poised-to-drive-160-increase-in-power-demand
3. https://about.bnef.com/blog/texas-tug-of-war-grid-juggles-solar-boom-and-power-hungry-data-centers/