AI Wants Power

The rapid development of artificial intelligence is fundamentally changing demand for electricity, transforming the energy sector into an unexpected but dynamic area of growth.

The unprecedented buildout of AI data center infrastructure is creating enormous opportunities for technology companies, especially those involved in semiconductors, networking, and specialized cooling solutions. The scale of investment is staggering. It is forecast that between 2026 and 2030, about 100 GW of new data centers will be built, translating into a $1.2 trillion increase in the value of real estate assets and an additional $1 to $2 trillion in IT equipment.

This infrastructure expansion of data centers is a direct driver of a boom for semiconductor manufacturers. Micron Technology (MU), a leader in memory and storage solutions, is
a major beneficiary. Micron’s products are essential for high-performance computing required by artificial intelligence (AI). The company’s financial results for fiscal year 2025 (TTM) are impressive. Net income rose by 997.6%, reflecting growing demand for components used in AI servers.
A P/E ratio (price to earnings) of 38.89 and a P/S ratio (price to sales) of 10.95 indicate strong market confidence in the company’s future growth trajectory.
Beyond chips, the physical design and equipment of AI data centers are also critical. Comfort Systems USA, Inc., an HVAC services provider, is positioned for significant growth. Currently priced at $1,337.95 with a market capitalization of $47.19 billion, Comfort Systems is a leader in designing and installing complex cooling systems required for high-density AI data centers. A single hyperscaler, meaning a data center that provides computing power through the cloud, can consume as much electricity as 100,000 households, and cooling systems draw millions of liters of water per day, making effective thermal management essential. This information was reported by Kavout,
a U.S. company offering advanced AI-based investment tools.

The effects of the AI technology boom will spread across the entire economy, far beyond technology stocks. This applies above all to energy companies that supply power to AI-oriented data centers. And because major technology companies, such as Microsoft and MSFT, plan to invest billions of dollars in building new data centers, AI-related power consumption is expected to rise, and to do so very rapidly.

Kinder Morgan (KMI) meets 20% of U.S. electricity demand and serves most of the major U.S. natural gas supply and demand regions. Stephen Ellis, an energy and utilities strategist at Morningstar, an investment research firm, says: “Kinder Morgan’s assets include natural gas, natural gas liquids, crude oil, and liquefied natural gas. The company’s U.S. pipeline operations are particularly impressive.” Morningstar cites Kinder Morgan management’s view that “Rising demand from artificial intelligence could increase gas demand from more than 2 billion cubic meters per day to more than 5 billion cubic meters per day by 2030. This surge in demand would mean AI energy demand of about 15% to 20% by 2030, up from 2.5% in 2022. Gas would cover about 40% of incremental demand, due to intermittency challenges with solar and wind power.”

Leaders in the AI industry, including Elon Musk and Andy Jassy of Amazon, have assessed that electricity supply is the main bottleneck for AI development,
a role previously played by limited chip availability.

Shane Neagle, Editor-in-Chief of The Tokenist, writes on Investing that the rapid development in artificial intelligence may be temporarily slowed in order to “catch up the logistics,” meaning to expand the energy sector to
a level that guarantees a steady and secure flow of power to “artificial brains.” The comparison is fitting, especially when we consider that the human brain consumes 20% to 25% of the body’s energy while accounting for only 2% of its mass. Data centers are also not giants in terms of physical footprint compared with other enterprises and institutions and other electricity users, yet their power needs are enormous. “Each query to GPT-4 uses nearly ten times more energy than a standard Google search (NASDAQ: GOOGL), and given billions of daily queries from millions of users, cumulative energy demand is reaching grid overload levels.” It can therefore be said to be approaching a critical level.

Shane Neagle notes: “This bottleneck is more than a temporary challenge. It is a structural shift requiring massive infrastructure investment over the next decade. Building new power plants, modernizing transmission grids, and connecting renewable energy sources are multi-year undertakings that cannot keep pace with AI’s exponential growth. The gap between energy demand and supply capacity is widening, creating a sustained opportunity for energy companies that can deliver reliable generation solutions and grid infrastructure.”

Many investors recognize that while the hype around artificial intelligence has driven technology valuations to extreme levels, real long-term value may be created in the energy sector, because only its expansion will enable continued AI development. With U.S. federal policy mobilizing to meet this challenge and corporations competing to secure energy supplies, energy stocks offer an attractive value proposition based on a fundamental imbalance between supply and demand that may persist for years.

Kavout reported in February of this year that “unlimited electricity demand from AI data centers is driving a fundamental transformation of the utilities sector from
a defensive area into a stable source of growth. Data center infrastructure providers and semiconductor companies are direct beneficiaries, but the primary bottleneck for expansion at this point is the physical availability of electricity at grid scale.”

The portal advises prospective investors to prioritize energy companies with strong balance sheets and a solid regulatory foundation, as well as infrastructure and semiconductor companies that offer innovative energy solutions or key components that enable those solutions.

Will electricity demand from artificial intelligence reshape the energy sector, Kavout asks, and answers: “The rapid development of artificial intelligence is fundamentally changing demand for electricity, transforming the energy sector into an unexpected but dynamic area of growth. This is not a temporary sentimental play. It is
a direct response to a fundamental, capital-intensive supercycle driven by the need to power advanced AI models and the massive data centers that house them.” Kavout also cites Morningstar forecasts that electricity demand in U.S. data centers is expected to triple between 2024 and 2030, ultimately consuming as much as 10% of the country’s total electricity. A single AI-related task can consume up to 1,000 times more electricity than traditional internet search, illustrating the immense computational intensity and energy demands of this new technological era.

The financial commitment of hyperscaler companies underscores this shift. As noted above, technology giants such as Meta, Microsoft, Amazon, and Alphabet are expected to allocate an astonishing $700 billion to AI development in 2026 alone. This structural impulse, rather than cyclical growth, represents a multi-year opportunity for companies able to provide the necessary power and infrastructure.

At present, the center of gravity has shifted from computing performance to the physical availability of electricity at grid scale, which makes “speed of power delivery” the most important factor determining project viability, Kavout writes.