Why I'm Betting on the Energy Market Due to AI's Electricity Bottleneck: My Two Cents
As someone who works in the AI space, I've witnessed firsthand the incredible advancements and the hype surrounding artificial intelligence. Companies like NVIDIA have seen their stock prices skyrocket to all-time highs, fueled by the insatiable demand for AI technologies. But beneath this excitement lies a critical, often overlooked issue that could significantly impact the energy sector—and offer intriguing investment opportunities.
In this article, I want to share my perspective on why I'm betting on the energy market, specifically utilities and natural gas companies, due to the emerging electricity bottleneck caused by AI's exponential growth. If you're an investor or just someone interested in the intersection of technology and energy, this is a trend you won't want to miss.
The Unseen Problem: Electricity Demand Outpacing Supply
The AI Boom and Its Energy Appetite
Working in AI, I see daily how models are becoming increasingly complex, requiring massive computational power to train and operate. While the parameter sizes of these models aren't doubling every few months, the growth is still substantial. For example, GPT-3 has 175 billion parameters, and estimates suggest GPT-4 has around 280 billion parameters (*1). Training GPT-4 is estimated to have required about 1,750 megawatt-hours of electricity—the equivalent of what 160 average American homes use in a year (*2).
But it's not just about training these models; running them (inference) also demands significant power. Each query to GPT-4 consumes about 2.9 watt-hours of electricity, nearly ten times that of a standard Google search (*3). Multiply that by millions of users and billions of queries, and you can see how quickly the energy consumption adds up.
Hitting the Limits of Electrical Infrastructure
Here's the crux of the issue: our current electrical infrastructure isn't equipped to handle the escalating demands of AI. Data centers already consume 1-2% of global electricity, and this figure is projected to rise to 3-4% by 2030 (*4). The International Energy Agency forecasts that global data center electricity demand will more than double from 2022 to 2026, with AI playing a major role (*5).
In my professional circles, there's growing concern about the strain on power infrastructure. Operating large clusters of high-performance GPUs, like NVIDIA's H100, could potentially strain a state's entire electrical grid. While specific figures vary, the general consensus is that we're nearing the limits of what our grids can handle (*6).
Microsoft seems to recognize this issue. They've recently purchased a power plant, presumably to secure a stable electricity supply for their data centers (*7). This move underscores the severity of the electricity bottleneck we're approaching.
The Impending Slowdown in AI Development
Given these constraints, I believe the rapid pace of AI advancement may slow down in the short to medium term. Industry leaders like Elon Musk and Amazon CEO Andy Jassy have identified electricity supply as the latest bottleneck for AI development, replacing the previous constraint of chip availability (*8). It's not just about technological capabilities anymore; it's about physical resources. We simply aren't producing enough electricity to sustain the current trajectory of AI scaling.
This isn't a hurdle we can clear overnight. Building new power plants, upgrading grid infrastructure, and securing renewable energy sources are massive undertakings that require time and substantial investment. This potential slowdown has significant implications for markets and investors, shifting attention toward sectors that can address or benefit from these challenges.
Why the Energy Sector Stands to Benefit
Increased Demand for Electricity
The most direct beneficiary of this situation is the energy sector. As AI companies grapple with electricity shortages, utilities and energy providers will see increased demand. According to Goldman Sachs Research, data center power demand is expected to grow 160% by 2030 (*4). This isn't just a temporary spike; it's a trend that could persist as long as the demand for AI technologies continues to grow.
Natural Gas as a Key Player
Natural gas is a cornerstone of U.S. electricity generation, accounting for approximately 43% of the country's electricity production in 2023 (*9). Its abundance, relatively low cost, and ability to quickly ramp up production make it essential for meeting immediate energy demands. With constraints on electricity supply, natural gas producers and related infrastructure companies are in a prime position to capitalize.
Opportunities in Grid Infrastructure
Beyond just producing more electricity, there's a pressing need to upgrade and expand the electrical grid. The strain isn't solely about capacity but also about managing fluctuations in demand. Companies specializing in grid infrastructure and smart technologies could see substantial growth as they help modernize the system to handle higher loads.
Stocks and Sectors I'm Watching
*Utilities with Natural Gas-Fired Power Plants *NextEra Energy (NEE): Not only does NextEra have significant natural gas operations, but they're also leaders in renewable energy. This dual focus positions them well for both immediate and long-term energy needs. ] *Duke Energy (DUK): Serving millions across multiple states, Duke Energy's extensive infrastructure makes them a key player in meeting increased electricity demand.
The information and publications are not meant to be, and do not constitute, financial, investment, trading, or other types of advice or recommendations supplied or endorsed by TradingView. Read more in the Terms of Use.