Electricity Is the New Oil | Rob Thummel on the Energy Opportunity from the AI Boom

Episode Summary

Executive Summary: The discussion argues that AI’s growth will be constrained by electricity, making power generation, grid expansion, and energy infrastructure the key investment themes. Rob Thummel says natural gas and nuclear will gain share, while U.S. shale has kept energy prices and inflation in check. He also highlights AI infrastructure beneficiaries beyond mega-cap tech, including data centers, cooling, and power-capacity plays.

Main Topics: Electricity as the constraint on AI growth (Priority: 5/5): Thummel argues that the pace of AI development will be limited less by chips than by access to electricity, making power availability the central bottleneck for hyperscalers and data centers. U.S. energy mix and expected shifts (Priority: 5/5): He outlines the current U.S. energy mix and expects natural gas and nuclear to gain share over the next several years, while coal likely declines and renewables remain important but limited by intermittency. Why electricity is the new oil (Priority: 5/5): The future economy is framed as AI-driven, and AI requires electricity. That makes electricity the essential input, similar to oil in the last century, for domestic and global growth. Renewables, storage, and grid limitations (Priority: 4/5): Wind and solar are described as useful but constrained by intermittency, transmission bottlenecks, and insufficient battery storage. A larger, more resilient grid is needed. Nuclear’s comeback and challenges (Priority: 4/5): Nuclear is seeing renewed interest due to carbon-free baseload power and direct demand from tech companies, but enrichment capacity and regulatory timelines remain major obstacles. U.S. shale’s macroeconomic impact (Priority: 5/5): Thummel credits shale technology for making the U.S. the largest oil and gas producer/exporter, reducing import dependence and helping keep inflation lower than it otherwise would have been. AI infrastructure investing opportunity (Priority: 4/5): Tortoise’s AI Infrastructure ETF focuses on enablers of AI: data centers, network hardware, cooling, power capacity, and related industrial players, many of which are not traditional energy companies.

Key Arguments: Electricity, not just semiconductors, is becoming the limiting factor for AI growth because hyperscalers need massive, reliable power access. Natural gas is the most practical near-term source of incremental electricity because it is abundant, relatively cheap, and lower-carbon than coal. Nuclear is regaining relevance because data centers need 24/7 baseload power, but the U.S. lacks enough uranium enrichment capacity to fully scale it quickly. Wind and solar alone cannot support AI/data center demand because they are intermittent and require more transmission and storage than currently exists. U.S. shale has been a major anti-inflation force by increasing domestic supply and making the U.S. a net energy exporter. Energy infrastructure can be a better investment than commodity producers because it is fee-based, volume-driven, and less sensitive to swings in oil and gas prices. AI creates second-order investment opportunities in cooling, storage, switches, fiber, and grid-capacity companies, not just in the largest AI platforms.

Data Points: U.S. electricity consumption: about 4,000 terawatt hours annually - Used as the baseline for current U.S. electricity demand. Projected U.S. electricity demand growth: about 3,000 more terawatt hours - Estimated additional electricity generation needed by 2050. Projected increase in U.S. electricity demand: 75% - Thummel’s estimate for demand growth between now and 2050. Equivalent added electricity: three Japans - He compares the needed incremental U.S. generation to Japan’s total electricity demand. Current U.S. energy mix: oil and natural gas each in the mid-30% range - He says oil and natural gas are the predominant U.S. primary energy sources. Wind and solar share of total U.S. energy demand: single digits - He characterizes wind and solar as a small portion of total U.S. energy demand. Nuclear share of total U.S. energy demand: around 10% - Approximate share when considering broader U.S. energy demand. Nuclear share of U.S. electricity generation: about 19% to 20% - Used to explain why uranium enrichment capacity matters. U.S. carbon emissions reduction from coal-to-gas switching: 20% since 2005 - He says natural gas replacing coal in power generation lowered emissions materially. Energy demand growth history: 40 of the last 42 years - Global energy demand has risen in nearly every year over the past four decades. Vacancy rate in existing data centers: 3% or below - Illustrates tight supply in the data center market. Stargate project size: $500 billion - A massive data center campus in Abilene, Texas, cited as an example of AI infrastructure buildout. Power footprint comparison for Meta’s Hyperion campus: size of Central Park - Used to show how large the planned data center campus is. Typical household electricity use: about 10,500 kilowatt-hours per year - Provided for context on individual household usage versus national demand. Energy infrastructure investor return profile: 5% dividend yields or more - He cites yields from energy infrastructure holdings as attractive to income investors. Energy infrastructure growth: 3% to 5% - Expected dividend/cash flow growth rate for infrastructure businesses. Bitcoin miner to data center conversion example: Riot Platforms, Terawulf, IREN - Companies he says are repurposing electricity capacity toward data center operations.

Pivotal Quotes: "Electricity is the new oil as the main driver of the future economy, both domestically and globally." — Rob Thummel: Explaining why AI and electricity will define future economic growth. "The best investments I've ever made are ones when everything seems like it's time not to be investing." — Rob Thummel: Closing lesson on contrarian investing and buying when sentiment is poor. "We're going to have to add three Japans of electricity here in the U.S. between now and 2050." — Rob Thummel: Illustrating the scale of additional power generation required for AI and economic growth.

Implications: Investors should focus on power, grid, and AI-enabling infrastructure rather than only on headline AI stocks. The next decade may reward firms tied to baseload generation, transmission, cooling, and fee-based energy transport as electricity becomes the key growth constraint.